CCSK Practice Test

Certificate of Cloud Security Knowledge (CCSK)

CCSK overview

  • 200 practice questions
  • 32 flashcards
  • 11 study guide topics
  • 40 key terms

What you'll study

  • Cloud Computing Concepts & Architectures9%
  • Cloud Governance9%
  • Risk, Audit, & Compliance9%
  • Organization Management9%
  • Identity & Access Management8%
  • Security Monitoring8%
  • Infrastructure & Networking8%
  • Cloud Workload Security8%
  • Data Security8%
  • Application Security8%
  • Incident Response & Resilience8%
  • Related Technologies & Strategies8%

The Certificate of Cloud Security Knowledge (CCSK) is a cloud security credential from the Cloud Security Alliance, described by CSA as the industry benchmark for cloud security expertise. CCSK v5 is aimed at IT professionals working in cloud computing, including security staff, developers, IT operations staff, audit and compliance professionals, sales and solution engineers, and product marketers. No official work experience or prior qualification is required, though a basic grasp of security fundamentals such as firewalls, secure development, encryption, and identity and access management helps. The exam is open-book, delivered online, and consists of 60 multiple-choice questions drawn randomly from a larger pool, to be completed in 120 minutes, with a minimum passing score of 80%. CCSK v5 covers 12 domains, from cloud computing concepts and architectures through governance, risk and compliance, identity and access management, security monitoring, infrastructure and networking, workload, data and application security, incident response and resilience, to related technologies and strategies. The curriculum adds emphasis on artificial intelligence and generative AI, Zero Trust, DevSecOps and CI/CD, automation, serverless workloads, and data lakes. Preparation options include a free CSA Prep Kit for self-study, online self-paced courses, and virtual or in-person instructor-led training, including a lectures-plus-labs format.

Common mistakes

When a stem names a specific component (guest OS, hypervisor, function runtime, object in a bucket), the candidate answers from the memorised slogan 'provider secures the cloud, customer secures what's in it' instead of resolving the split for the service model described - for example saying the provider patches the guest OS of an IaaS VM, or that the customer configures the hypervisor.

The slogan is the most repeated line in cloud security and is only true at its original level of abstraction; on-premises habits and consumer-cloud experience both push toward a fixed split, and candidates rarely practise resolving it component by component.

What to do instead

Before reading the options, write down (a) the deployment model, (b) the service model, (c) the exact component named, then assign it. Memorise the invariants: data classification, identity, access policy and tenant configuration stay customer tasks in every model; the guest OS is customer-patched in IaaS but provider-patched in PaaS and SaaS; the hypervisor is always the provider's.

Treating CSA's publications as interchangeable - describing the CAIQ as an audit report or attestation, calling the CCM a certification, or asserting that STAR Level 1 means a third-party-audited provider.

CCM, CAIQ, STAR, the Security Guidance and ISO/IEC 27017/27018 all sit under the same CSA banner and appear in the same paragraphs of study material, so they collapse into one blurry 'CSA compliance thing'.

What to do instead

Anchor one sentence per artifact: Security Guidance = best-practice document organised into domains; CCM = catalogue of control specifications across 17 control domains; CAIQ = self-assessment questionnaire whose answers map to CCM controls; STAR = programme in which Level 1 is self-assessment, Level 2 is a third-party audit and Level 3 is continuous assessment; ISO/IEC 27017 and 27018 are certifications, not CSA documents.

Answering governance questions with a compliance artifact - choosing 'obtain the provider's SOC 2 report' or 'map the service to ISO 27001' when the stem is really asking who decides, who is accountable, or how cloud is structured inside the organisation.

Governance and compliance are used interchangeably in vendor marketing, and auditable documents are the easiest things to name, so they become the default answer whenever the word 'governance' appears.

What to do instead

Separate the concepts: governance = decision rights and accountability, the policy hierarchy (policy, standard, procedure, guideline), defined roles such as cloud service owner and cloud security architect, plus portfolio and provider management; compliance = evidence that requirements are met. Keep provider reports for options about assurance evidence and look for a policy, role or ownership answer when the stem talks about authority or decision-making.

Using OAuth 2.0 and OpenID Connect interchangeably for authentication, or saying SAML issues access tokens for API authorisation, when asked how a user or workload is authenticated or how delegated access is granted.

Sign-in buttons are commonly labelled 'Log in with X, powered by OAuth', and all three protocols appear together in federation discussions, so the distinction between proving identity and granting delegated access is lost.

What to do instead

Hold three one-liners: SAML 2.0 = XML assertions for browser-based SSO and federation; OAuth 2.0 = delegated authorisation that issues access tokens and does not authenticate the user; OpenID Connect = identity layer on top of OAuth 2.0 that adds an ID token (JWT) and authenticates the user. Add SCIM for cross-domain identity provisioning, and remember that federation is not provisioning.

Assuming the provider's native control-plane logs are the customer's complete monitoring story - answering that no additional logging is needed for an IaaS workload, or that the provider is responsible for detecting attacks on the customer's application.

Provider consoles present rich dashboards and 'the provider logs everything' is a comfortable assumption; candidates underestimate that provider logs are scoped to the provider's own plane and that the customer must both enable and export them.

What to do instead

Split monitoring by layer and state who produces each: the provider supplies control-plane/API audit logs and infrastructure telemetry; the customer enables them, adds OS, application, data-access and network-flow logging, centralises into a SIEM, writes detections and alerting, and protects log integrity and retention. Remember logs are customer data and must be exported and retained independently of the workload's lifecycle.

Mixing up stateful instance-level firewalling with stateless subnet-level filtering (security groups vs network ACLs), placing a WAF where a network control is wanted, or stating that VPC/VNet peering or a private endpoint carries traffic over the public internet.

The controls overlap in purpose, each provider renames them, and 'private connection' sounds like a tunnel over the internet to anyone who has not drawn the actual packet path.

What to do instead

Learn properties, not brand names: stateful, allow-only rules attached to an instance/ENI versus stateless rules supporting allow and deny attached to a subnet; WAF is application-layer HTTP content inspection and no substitute for network filtering. For connectivity, know that peering and private endpoints keep traffic on the provider backbone and off the internet, and that peering is non-transitive so a transit hub is needed for multi-site meshes.

Applying VM patching logic to containers and functions - saying the customer patches the host OS of a managed container service, or that the provider is responsible for a function's third-party libraries and its execution role.

'Workload' is treated as a single category, and the shared-responsibility question is answered on autopilot from the IaaS case the candidate studied most.

What to do instead

Trace the boundary explicitly: for managed orchestration the provider patches the control plane and, on managed nodes, the host OS, while the customer owns the image, base-image currency, application dependencies and cluster/RBAC configuration; with self-managed nodes the customer also owns the host OS. For functions the provider owns the runtime and physical host, while the customer owns the code, dependencies, timeout/memory configuration and the execution role's permissions.

Answering that data is 'protected because it is encrypted at rest' without asking who holds the keys and who can reach plaintext, and using DLP, CASB, tokenisation and encryption as if they were synonyms.

Encryption is taught as a checkbox, so the existence of ciphertext gets equated with control of the data, and key custody, key residency and provider administrative access never enter the reasoning.

What to do instead

For each data question resolve three things separately: where the ciphertext lives, who controls the keys (provider-managed, customer-managed KMS, BYOK/HYOK, client-side), and who can obtain plaintext (provider operators, support access, a compromised tenant administrator). Keep the tool definitions distinct: encryption renders data unreadable; tokenisation substitutes non-sensitive surrogates with the mapping held elsewhere; DLP detects or masks sensitive content; CASB is a policy enforcement point between users and cloud services; key management is a separate control with its own availability risk.

Matching the wrong testing technique to the stem - selecting DAST when the question describes reviewing source or open-source dependencies, or claiming SAST finds runtime configuration errors in a deployed environment.

SAST, DAST, SCA, IAST and RASP are acronyms memorised as one table, and exam stems describe the activity rather than naming the acronym, so pattern-matching on keywords fails.

What to do instead

Map technique to what it can actually see: SAST reads source or bytecode without executing (white box, early, higher false positives); DAST probes a running application from outside (black box, no source needed); SCA inventories third-party libraries and licences; IAST instruments the running application from inside; RASP blocks at runtime in production. Dependency or licence language means SCA; a live URL with no source access means DAST.

Applying on-premises incident response mechanics in the cloud - proposing to power off or reimage a suspected instance as the first containment step, assuming disk imaging is the only evidence method, or assuming the provider will investigate the customer's workload for them.

'Pull the plug and take the disk' is deeply practised, the loss of volatile state, instance-store data and log streams on termination is not intuitive, and the provider's scale creates a false expectation of managed forensics.

What to do instead

Adapt the phases: preserve first by snapshotting volumes, capturing memory where supported, exporting and copying logs into a separate account, and recording instance metadata and security-group state; then contain through the API (revoke and rotate credentials, remove the instance from the network via security groups or a quarantine subnet) rather than destroying evidence; rebuild from known-good images instead of cleaning in place. Know the boundary - the provider secures its own infrastructure and may give log access and support, but investigating the customer's workload and meeting notification obligations remain the customer's. For resilience questions, keep RTO (tolerable downtime) and RPO (tolerable data loss) explicit and remember multi-AZ availability is not disaster recovery.

Exam requirements can change. Confirm current details with the exam provider.

CCSK practice

200 questions. Each one lists the choices, the correct answer, and an explanation. With JavaScript, this list opens one question at a time. The checker at the top of this page saves progress in this browser when it is available.

  1. Question 1Cloud Computing Concepts & Architectures

    Which NIST cloud characteristic lets a consumer provision server time and network storage unilaterally, without human interaction with the provider?

    1. On-demand self-service Correct answer.
    2. Measured, metered service
    3. Broad network access
    4. Multi-tenant resource pooling

    Explanation

    NIST defines on-demand self-service as the ability of a consumer to unilaterally provision computing capabilities, such as server time and network storage, automatically and without requiring human interaction with each service provider. The remaining characteristics describe how services are reached, how resources are shared, and how usage is metered rather than how provisioning autonomy works.

  2. Question 2Cloud Computing Concepts & Architectures

    In which service model does the provider manage the application itself while the consumer simply uses it, with no management of the underlying infrastructure?

    1. SaaS Correct answer.
    2. FaaS
    3. IaaS
    4. PaaS

    Explanation

    In Software as a Service the provider runs and maintains the entire stack including the application, and the consumer accesses it through a client interface without managing servers, platforms or application code. The other models expose progressively more of the stack to the consumer for management.

  3. Question 3Cloud Computing Concepts & Architectures

    Which deployment model describes cloud infrastructure provisioned for exclusive use by a single organization comprising multiple consumers, such as separate business units?

    1. Private cloud Correct answer.
    2. Public cloud
    3. Hybrid cloud
    4. Community cloud

    Explanation

    A private cloud is provisioned for exclusive use by a single organization with multiple consumers, and it may be owned, managed and operated by the organization or a third party on or off its premises. Community clouds serve several organizations with shared concerns, while public clouds are open to the general public.

  4. Question 4Cloud Computing Concepts & Architectures

    Which component runs directly on the physical host hardware without requiring an underlying general-purpose operating system?

    1. Type 1 hypervisor Correct answer.
    2. Paravirtualized guest OS
    3. Hosted hypervisor
    4. Type 2 hypervisor

    Explanation

    A Type 1, or bare-metal, hypervisor is installed directly on the host hardware and does not depend on a general-purpose host operating system, which reduces the attack surface and often improves performance. Type 2 hypervisors run as applications on a conventional operating system.

  5. Question 5Cloud Computing Concepts & Architectures

    Which cloud storage model is best suited to large volumes of unstructured data accessed over HTTP APIs through a flat namespace of unique keys?

    1. Block storage
    2. Direct-attached storage
    3. File storage
    4. Object storage Correct answer.

    Explanation

    Object storage keeps data as discrete objects with unique keys and rich metadata in a flat namespace, reached through RESTful HTTP APIs, which makes it well suited to massive unstructured datasets and web-scale access. Block and file storage present different abstractions intended for different workloads.

  6. Question 6Cloud Computing Concepts & Architectures

    Under the shared responsibility model for an IaaS deployment, which task remains the customer's responsibility?

    1. Patching the guest operating system Correct answer.
    2. Securing the physical network fabric
    3. Patching the hypervisor
    4. Maintaining the physical data center

    Explanation

    In IaaS the provider secures and operates the facilities, hardware, network fabric and hypervisor, while the customer owns everything from the guest operating system upward, including its patches, configuration and applications. The boundary shifts upward as customers adopt PaaS and SaaS.

  7. Question 7Cloud Computing Concepts & Architectures

    Which practice describes running an application primarily in a private environment and automatically extending capacity into a public cloud when demand peaks?

    1. Moving peak overflow demand from a private environment into a public cloud Correct answer.
    2. Migrating all workloads permanently from private to public cloud
    3. Balancing traffic across availability zones within a single region
    4. Running workloads across two public providers for redundancy

    Explanation

    Cloud bursting keeps the baseline workload in a private or dedicated environment and temporarily adds public cloud capacity when demand exceeds local resources, after which the overflow is released. It is a capacity strategy, not a permanent migration or an availability-zone design pattern.

  8. Question 8Cloud Computing Concepts & Architectures

    What primarily distinguishes container-based workloads from virtual machines?

    1. Containers cannot run on Linux hosts
    2. Containers require a guest OS per instance
    3. Containers provide stronger hardware isolation than virtual machines
    4. Containers share the host kernel instead of virtualizing hardware Correct answer.

    Explanation

    Containers package an application with its dependencies and rely on operating-system-level isolation, sharing the host kernel rather than each carrying a full guest operating system. This makes them lighter and faster to start, but it also means the shared kernel becomes part of the trust boundary.

  9. Question 9Cloud Computing Concepts & Architectures

    A company keeps regulated records in its own data center while running its web front end in a public cloud, with orchestration linking the two. Which deployment model does this illustrate?

    1. Hybrid cloud Correct answer.
    2. Community cloud
    3. Private cloud
    4. Public cloud

    Explanation

    A hybrid cloud combines two or more distinct deployment models that remain unique entities but are bound together by standardized or proprietary technology enabling data and application portability. The split between an on-premises estate and a public provider, joined by orchestration, is the defining pattern.

  10. Question 10Cloud Computing Concepts & Architectures

    Which term describes the ability to move application workloads and data from one cloud provider to another with minimal disruption and rework?

    1. Federation
    2. Elasticity
    3. Portability Correct answer.
    4. Interoperability

    Explanation

    Portability is the capability to move components such as applications and data between providers or environments without significant re-engineering. Interoperability is the related but distinct ability of systems to exchange and use information, which does not by itself permit migration.

  11. Question 11Cloud Computing Concepts & Architectures

    Which mechanism lets a user authenticated by the home organization's identity provider reach a cloud service without a separate set of cloud-specific credentials?

    1. Multi-factor authentication
    2. Federated identity Correct answer.
    3. Mandatory access control
    4. Role-based access control

    Explanation

    Federated identity establishes a trust relationship between the user's home identity provider and the cloud service provider, so assertions about the authenticated user are accepted across organizational boundaries. This avoids maintaining parallel local credentials at each provider.

  12. Question 12Cloud Computing Concepts & Architectures

    How does a cloud provider typically structure regions and availability zones to support resiliency?

    1. Each region is limited to exactly one availability zone
    2. A region contains multiple isolated zones with independent power and network Correct answer.
    3. All availability zones share one power and cooling infrastructure
    4. Availability zones are logical partitions inside a single data hall

    Explanation

    A region is a geographic area containing multiple availability zones that are physically separated with independent power, cooling and network connectivity, so a failure in one zone does not cascade to others. Customers achieve resilience by deploying redundant instances across zones within a region.

  13. Question 13Cloud Computing Concepts & Architectures

    Which cloud model runs individual code units on demand, scales automatically, and bills only for the execution time consumed while the provider manages all server infrastructure?

    1. Bare-metal hosting
    2. IaaS with reserved instances
    3. Managed Kubernetes clusters
    4. Serverless compute Correct answer.

    Explanation

    Serverless compute, typically delivered as Functions as a Service, abstracts servers entirely: the provider provisions capacity, scales on demand and charges per invocation and execution duration. The customer supplies code and configuration only, never host or cluster management.

  14. Question 14Cloud Computing Concepts & Architectures

    What is a defining feature of software-defined networking in cloud environments?

    1. Network functions run only on dedicated hardware appliances
    2. Control plane is decoupled from the data plane and managed programmatically Correct answer.
    3. Network topology is fixed at provisioning time
    4. Traffic is routed solely by physical switch MAC tables

    Explanation

    Software-defined networking separates the control plane that decides where traffic goes from the data plane that forwards it, exposing control through APIs so networks can be defined in software. This is what allows cloud providers and tenants to create isolated virtual networks, subnets and security policies on demand.

  15. Question 15Cloud Computing Concepts & Architectures

    Why is destroying the encryption keys often the preferred way to render deleted cloud data unrecoverable?

    1. Compressing data reduces the storage cost of deletion
    2. It proves the provider cannot access the data
    3. Destroying the keys renders the remaining ciphertext unrecoverable Correct answer.
    4. It automatically replicates data to a second region

    Explanation

    Cryptographic erasure leaves ciphertext in place but destroys the key material needed to decrypt it, which is practical in multi-tenant environments where the exact physical media location of the data may be unknown or replicated. The technique depends on effective key management and key destruction being auditable.

  16. Question 16Cloud Computing Concepts & Architectures

    Which risk is most closely associated with co-locating virtual machines from different tenants on the same hypervisor?

    1. Increased latency from network oversubscription
    2. Cross-tenant compromise through a hypervisor or side-channel exploit Correct answer.
    3. Loss of the ability to encrypt data at rest
    4. Reduced elasticity during demand spikes

    Explanation

    Multi-tenancy places logically isolated instances on shared physical hosts, so a hypervisor vulnerability or a cache-timing side channel can become a path from one tenant's workload to another's data. This is why providers emphasize hypervisor patching, hardware-assisted isolation and dedicated-host options for the most sensitive workloads.

  17. Question 17Cloud Computing Concepts & Architectures

    What is the principal security consequence of cloud platforms being API-driven and remotely managed?

    1. Encryption at rest becomes unnecessary
    2. API access cannot be logged or audited
    3. APIs remove the need for authentication
    4. Compromise of management APIs can expose entire environments Correct answer.

    Explanation

    Because provisioning, configuration and data access are performed through management APIs, a stolen or over-privileged key can act at machine speed across the whole estate. This is why API credential protection, least privilege, key rotation and management-plane logging are central cloud controls.

  18. Question 18Cloud Computing Concepts & Architectures

    A provider automatically replicates customer data into several geographic regions for durability. What is the primary governance concern this creates?

    1. Data residency and legal jurisdiction of stored copies Correct answer.
    2. Increased storage cost only
    3. Loss of data durability
    4. Reduced availability during outages

    Explanation

    Cross-region replication can place copies of data in other countries or legal jurisdictions, which may conflict with residency requirements, contractual commitments or data protection law. Organizations must map where data is stored and processed, and confirm which replication features can be constrained.

  19. Question 19Cloud Governance

    In cloud governance, which statement best distinguishes governance from management?

    1. Governance applies only to financial reporting; management covers security operations.
    2. Governance executes controls; management defines the risk appetite.
    3. Governance is performed only by the cloud provider; management only by the customer.
    4. Governance sets strategic direction and oversight; management executes day-to-day activities. Correct answer.

    Explanation

    Governance establishes direction, priorities, risk appetite, and oversight through leadership and policy. Management carries out the operational activities that implement that direction. The two are complementary but not interchangeable.

  20. Question 20Cloud Governance

    An organization migrates to a SaaS provider. Which governance responsibility remains with the customer?

    1. Maintaining the physical security of the provider's data centers
    2. Defining which data may be stored and who may access it Correct answer.
    3. Patching the provider's hypervisor and host operating systems
    4. Managing the provider's internal employee background checks

    Explanation

    Under the shared responsibility model, the customer retains accountability for its own data, including classification, permitted use, and access decisions. Physical security and provider platform patching generally belong to the provider in a SaaS arrangement. The customer cannot delegate its data governance obligations to the provider.

  21. Question 21Cloud Governance

    Which sequence correctly orders the typical hierarchy of security documents from highest to most specific?

    1. Standard, policy, procedure, guideline
    2. Procedure, policy, standard, guideline
    3. Guideline, procedure, standard, policy
    4. Policy, standard, procedure, guideline Correct answer.

    Explanation

    Policy states management intent and mandatory high-level requirements. Standards define specific mandatory rules, procedures describe how to perform tasks, and guidelines provide recommended practices. This ordering moves from broad direction to detailed implementation guidance.

  22. Question 22Cloud Governance

    What does risk appetite represent in an enterprise risk management program?

    1. The amount and type of risk an organization is willing to retain Correct answer.
    2. The technical controls used to reduce identified vulnerabilities
    3. The maximum financial loss a single system can tolerate
    4. The list of risks that have been transferred to an insurer

    Explanation

    Risk appetite is the level and type of risk an organization is willing to accept in pursuit of its objectives. It guides decisions about which risks to avoid, mitigate, transfer, or accept. It is distinct from risk tolerance, which sets acceptable variation around that appetite.

  23. Question 23Cloud Governance

    What is the primary purpose of the CSA Cloud Controls Matrix (CCM)?

    1. To replace an organization's enterprise risk management program
    2. To measure the latency and availability of cloud data centers
    3. To provide a controls framework mapped to multiple standards Correct answer.
    4. To certify cloud providers against a single global privacy law

    Explanation

    The Cloud Controls Matrix provides a cloud-specific set of security controls and maps them to various standards and regulations. Organizations use it to assess providers and align cloud controls with broader compliance requirements. It is a controls framework, not a certification or performance measurement tool.

  24. Question 24Cloud Governance

    A company wants assurance that it can examine a cloud provider's security controls and records. Which contractual element best supports this?

    1. A right-to-audit clause covering relevant controls Correct answer.
    2. A service level agreement that lists uptime percentages
    3. A data residency clause naming the primary region
    4. A limitation-of-liability clause capping damages

    Explanation

    A right-to-audit clause grants the customer or its designee the ability to review the provider's controls, records, and audit reports. It supports ongoing governance and third-party risk oversight. Uptime, residency, and liability terms address different concerns.

  25. Question 25Cloud Governance

    What is the main role of COBIT in cloud governance?

    1. It provides an enterprise IT governance framework Correct answer.
    2. It certifies individual cloud data centers for physical security
    3. It defines technical configurations for cloud firewalls
    4. It prescribes cryptographic algorithms for data at rest

    Explanation

    COBIT is a framework for the governance and management of enterprise IT. It helps organizations define objectives, assign accountability, and assess IT-related risk and performance, including cloud services. It does not prescribe low-level technical configurations or certify facilities.

  26. Question 26Cloud Governance

    A risk assessment finds that a legacy application cannot be secured in the cloud. The organization retires the application instead of migrating it. Which risk treatment is this?

    1. Risk mitigation
    2. Risk avoidance Correct answer.
    3. Risk transfer
    4. Risk acceptance

    Explanation

    Risk avoidance eliminates the activity or asset that creates the risk, which is what retiring the application does. Risk mitigation reduces risk through controls, transfer shifts financial impact, and acceptance retains the risk knowingly. Retiring the service removes the exposure entirely.

  27. Question 27Cloud Governance

    What does a SOC 2 Type II report primarily provide to a cloud customer?

    1. An auditor's opinion on control effectiveness over a period Correct answer.
    2. A complete list of the provider's source code vulnerabilities
    3. A guarantee that no security incidents will occur in the future
    4. A certification of compliance with every global privacy regulation

    Explanation

    A SOC 2 Type II report covers the design and operating effectiveness of controls over a period of time, based on an independent auditor's testing. It gives customers evidence about controls but does not guarantee future outcomes. It is also not a complete vulnerability list or universal privacy certification.

  28. Question 28Cloud Governance

    In information governance, which role typically decides who may access a dataset and for what purposes?

    1. The data owner, who is accountable for the data Correct answer.
    2. The data custodian, who operates the storage platform
    3. The external auditor performing the annual review
    4. The cloud provider's facilities manager

    Explanation

    The data owner is accountable for a dataset and typically sets access rules and permitted uses. The custodian implements and operates the technical controls on the owner's behalf. Auditors provide independent assessment but do not set access policy.

  29. Question 29Cloud Governance

    Which governance approach best addresses employees adopting unsanctioned cloud services?

    1. Blocking all outbound traffic permanently without exception
    2. Delegating all cloud purchasing to individual business units
    3. Maintaining a sanctioned service catalog with light approval Correct answer.
    4. Relying on the finance department to detect recurring charges

    Explanation

    Shadow IT is best managed by making approved services easy to adopt while still applying risk review. A sanctioned catalog with a lightweight approval path gives users a compliant alternative. Blanket blocking and after-the-fact expense detection do not address the underlying business need.

  30. Question 30Cloud Governance

    Why is continuous compliance monitoring especially important in cloud environments?

    1. Cloud configurations change frequently, making audits quickly stale Correct answer.
    2. Continuous monitoring eliminates the need for any internal controls
    3. Cloud providers are legally required to publish all audit findings
    4. Point-in-time audits are prohibited by most standards

    Explanation

    Cloud resources are provisioned and changed rapidly, so a configuration that was compliant during an annual audit may drift soon afterward. Continuous monitoring detects drift and control failures closer to real time. It complements, rather than replaces, internal controls and periodic audits.

  31. Question 31Cloud Governance

    An enterprise uses several cloud providers with different control frameworks. Which practice most improves governance consistency?

    1. Accepting each provider's self-attestation without review
    2. Standardizing on the provider with the lowest cost
    3. Mapping provider controls to a single internal control framework Correct answer.
    4. Letting each business unit define its own unrelated controls

    Explanation

    A single internal control framework provides a common language for comparing providers and reporting risk. Provider-specific controls can be mapped into that framework so gaps and overlaps are visible. Decentralized or unverified approaches reduce comparability and oversight.

  32. Question 32Cloud Governance

    A team argues that moving workloads to a private cloud eliminates the need for cloud risk assessment. What is the most accurate governance response?

    1. Disagree, because private cloud still carries shared and operational risks Correct answer.
    2. Disagree, because private clouds are exempt from compliance requirements
    3. Agree, because risk assessment is only required for public cloud services
    4. Agree, because private clouds are always operated by internal staff

    Explanation

    Private cloud still involves shared infrastructure, operational dependencies, and concentration risk, even when operated internally. Governance requires risk assessment for any cloud deployment model. The deployment model changes the risk profile but does not remove the need to assess it.

  33. Question 33Cloud Governance

    Which set of measures best demonstrates cloud governance effectiveness to executive management?

    1. Counts of firewall rules and server names per region
    2. Number of cloud providers evaluated and subscription costs
    3. Asset ownership coverage and timely exception remediation Correct answer.
    4. Total employees attending any security awareness session

    Explanation

    Effective governance metrics should show accountability and follow-through, such as how much of the cloud estate has assigned owners and whether exceptions are remediated on time. Raw technical counts and attendance figures do not demonstrate governance outcomes. Executive reporting should focus on risk and control effectiveness.

  34. Question 34Cloud Governance

    Which governance control best addresses the risk of excessive entitlements accumulating in cloud identities over time?

    1. Increasing password complexity requirements for all users
    2. Requiring users to acknowledge an acceptable use policy annually
    3. Enabling static, long-lived access keys for service accounts
    4. Periodic access reviews with entitlement recertification Correct answer.

    Explanation

    Entitlement creep occurs when users or services retain permissions beyond what their current role requires. Periodic access reviews require managers or resource owners to recertify access and remove unnecessary rights. Password rules and policy acknowledgments do not address accumulated permissions.

  35. Question 35Cloud Governance

    What is the purpose of a cloud governance steering committee?

    1. To negotiate individual employee salaries in the cloud team
    2. To provide cross-functional oversight of cloud strategy and risk Correct answer.
    3. To write application code for cloud-native services
    4. To perform daily patch management across all cloud servers

    Explanation

    A governance steering committee brings together business, security, risk, and IT stakeholders to set cloud direction, resolve conflicts, and oversee risk and policy. It is an oversight body rather than an operational team. Technical execution remains with the appropriate engineering and operations groups.

  36. Question 36Cloud Governance

    Which statement best describes the relationship between compliance and governance in cloud computing?

    1. Governance gives oversight and direction; compliance shows adherence Correct answer.
    2. Compliance guarantees that governance objectives are fully met
    3. Compliance is optional when governance is mature
    4. Governance and compliance are identical and interchangeable terms

    Explanation

    Governance sets direction, accountability, and risk appetite, while compliance demonstrates that obligations and controls are being met. They are related but distinct: strong governance drives compliance activities, and compliance evidence informs governance. Neither replaces the other.

  37. Question 37Risk, Audit, & Compliance

    Which CSA artifact is a structured framework of security controls, organized by domain and mapped to major standards, that organizations use to assess a cloud provider's control environment?

    1. The Cloud Controls Matrix (CCM) Correct answer.
    2. The Top Threats to Cloud Computing report
    3. The Consensus Assessments Initiative Questionnaire (CAIQ)
    4. The Security, Trust, Assurance and Risk (STAR) registry

    Explanation

    The CCM is CSA's controls framework: it catalogues security controls by domain and maps them to standards such as ISO/IEC 27001, COBIT and PCI DSS so that cloud environments can be assessed consistently. The CAIQ is a questionnaire built on top of the CCM, while STAR is a programme that publishes assessments.

  38. Question 38Risk, Audit, & Compliance

    What is the primary purpose of the CSA Consensus Assessments Initiative Questionnaire (CAIQ)?

    1. To define the legal terms that govern cloud service contracts
    2. To certify a provider's infrastructure against the ISO/IEC 27001 standard
    3. To document provider controls using a standard set of questions Correct answer.
    4. To benchmark provider pricing against industry averages

    Explanation

    The CAIQ gives customers a consistent, industry-agreed set of yes/no questions so that provider responses can be compared across vendors and mapped back to the CCM. It is a disclosure instrument that supports due diligence, not a certification or a contract template.

  39. Question 39Risk, Audit, & Compliance

    A cloud provider wants to show that its controls have been examined by an independent third-party assessor, with the results published in the CSA STAR registry. Which STAR level applies?

    1. Level 1: self-assessment
    2. Level 2: third-party audit Correct answer.
    3. Level 3: continuous monitoring-based certification
    4. Level 4: government accreditation

    Explanation

    STAR Level 2 requires an independent third-party assessment of the provider's controls against the CCM or a recognized standard, and the resulting report is published in the registry. Level 1 is a self-assessment, and Level 3 adds continuous monitoring-based certification on top of an audited assessment.

  40. Question 40Risk, Audit, & Compliance

    Which standard supplies cloud-specific controls and implementation guidance that supplement ISO/IEC 27001 and ISO/IEC 27002?

    1. ISO/IEC 27036
    2. ISO/IEC 27018
    3. ISO/IEC 27017 Correct answer.
    4. ISO/IEC 27701

    Explanation

    ISO/IEC 27017 extends the ISO/IEC 27001 and 27002 control sets with cloud-specific guidance, and it explicitly clarifies which controls apply to cloud service providers and which to cloud customers. This makes it the reference point for dividing security responsibilities in a cloud relationship.

  41. Question 41Risk, Audit, & Compliance

    ISO/IEC 27018 is best described as which of the following?

    1. A certification scheme for the physical security of cloud data centres
    2. A code of practice for protecting PII in public cloud services Correct answer.
    3. A standard for cryptographic key management in the cloud
    4. A framework for cloud service level agreement metrics

    Explanation

    ISO/IEC 27018 is a code of practice that applies to public cloud providers acting as processors of personally identifiable information, adding requirements on consent, disclosure to third parties, and return or deletion of PII. It is used alongside ISO/IEC 27001 and 27017 rather than replacing them.

  42. Question 42Risk, Audit, & Compliance

    Which assurance report is intended for general distribution and summarizes a service organization's controls without disclosing detailed test results and exceptions?

    1. SOC 3 Correct answer.
    2. SOC 1 Type 1
    3. SOC 1 Type 2
    4. SOC 2 Type 2

    Explanation

    A SOC 3 report is a general-use report that gives the auditor's opinion on whether the service organization's description is fairly presented and its controls suitably designed, without the detailed control testing results included in SOC 2 reports. SOC 2 reports are restricted-use and distributed under confidentiality to specified parties.

  43. Question 43Risk, Audit, & Compliance

    In an IaaS deployment, a customer's auditor asks for evidence that guest operating system patches on the customer's virtual machines were applied within policy. Who must normally supply that evidence?

    1. The auditor, by scanning the provider's infrastructure
    2. The customer, because guest OS patching is a customer responsibility in IaaS Correct answer.
    3. The provider, because the hypervisor is in scope of its certification
    4. A third-party assessor, because patch evidence can only be validated externally

    Explanation

    In IaaS the provider secures the facilities, hardware, network and virtualization layer, while the customer controls the guest operating system, middleware and applications, including patching. The provider's audit reports therefore cannot evidence guest OS patch levels on customer-controlled instances.

  44. Question 44Risk, Audit, & Compliance

    An organization buys cyber insurance to cover losses arising from a cloud data breach. Which risk treatment option does this represent?

    1. Risk transference Correct answer.
    2. Risk mitigation
    3. Risk acceptance
    4. Risk avoidance

    Explanation

    Risk transference shifts the financial consequence of a risk to another party, typically through insurance, indemnities or contractual terms; the underlying risk event can still occur. Avoidance eliminates the activity, mitigation reduces likelihood or impact, and acceptance retains the exposure.

  45. Question 45Risk, Audit, & Compliance

    After controls have been implemented, the level of risk that still remains is known as which of the following?

    1. Inherent risk
    2. Risk tolerance
    3. Residual risk Correct answer.
    4. Risk appetite

    Explanation

    Residual risk is the exposure that persists once existing controls have been applied and evaluated. It is compared against the organization's risk appetite to decide whether further treatment is required.

  46. Question 46Risk, Audit, & Compliance

    Which of the following best defines risk appetite?

    1. The variance between assessed and realized risk
    2. The set of risks recorded in the risk register
    3. The amount and type of risk an organization is willing to accept Correct answer.
    4. The maximum financial loss that a single control failure could cause the organization

    Explanation

    Risk appetite expresses the amount and type of risk an organization is willing to accept in pursuit of its objectives, and it sets the boundary against which residual risk is judged. Tolerance then defines the acceptable variation around that appetite for specific metrics.

  47. Question 47Risk, Audit, & Compliance

    Before signing with a new SaaS provider, a company reviews the provider's SOC 2 report, CAIQ responses and public breach history. This activity is best described as which of the following?

    1. Continuous control monitoring
    2. Third-party risk due diligence Correct answer.
    3. Penetration testing
    4. Data loss prevention

    Explanation

    Due diligence is the pre-contract assessment of a prospective provider's security, privacy and compliance posture, drawing on independent audit reports, standardized questionnaires such as the CAIQ, and incident history. Its outputs feed the risk decision and the contractual requirements imposed on the provider.

  48. Question 48Risk, Audit, & Compliance

    A customer's contract with a large public cloud provider contains no right-to-audit clause. What is the most practical way for the customer to obtain assurance over the provider's controls?

    1. Commission an unannounced penetration test against the provider's production infrastructure
    2. Rely on the provider's independent audit reports and certifications Correct answer.
    3. Insist on an on-site inspection, since cloud customers always retain an unconditional right to audit provider facilities
    4. Accept the provider's public marketing statements about security as sufficient assurance

    Explanation

    Customers of hyperscale providers generally cannot obtain on-site audit rights; instead they rely on independent third-party reports and certifications such as SOC 2, ISO/IEC 27001 or STAR, then map those results to their own requirements and supplement them with contractual commitments. Unauthorized testing of provider infrastructure breaches the terms of service and may be unlawful.

  49. Question 49Risk, Audit, & Compliance

    Which approach best maintains assurance over a cloud provider's controls between periodic third-party audits?

    1. Replacing the auditor with an internal self-attestation each quarter
    2. Continuous monitoring of the provider's control evidence Correct answer.
    3. Relying on the original audit report until its expiry date
    4. Requesting a new full-scope audit whenever a configuration change occurs

    Explanation

    Continuous monitoring collects control evidence on an ongoing basis, such as configuration data, logs and updated attestations, so assurance reflects current conditions rather than a point in time. Re-auditing after every change is disproportionate, and a prior report's findings do not remain valid indefinitely.

  50. Question 50Risk, Audit, & Compliance

    A multinational must keep EU personal data within the EU. Which contractual and architectural element most directly addresses this requirement?

    1. Contractual and technical guarantees that data stays in EU regions Correct answer.
    2. A service level agreement guaranteeing 99.99% availability
    3. A right-to-audit clause covering the provider's global data centres
    4. A clause requiring encryption of data at rest with customer-managed keys

    Explanation

    Data residency is achieved through contractual commitments combined with technical controls such as region pinning, restricted replication and configuration that confines storage and processing to approved jurisdictions. Encryption protects confidentiality but does not by itself prevent data from being stored or processed outside the EU.

  51. Question 51Risk, Audit, & Compliance

    A cloud customer receives a litigation hold covering data held in a SaaS application. Which activity primarily rests with the customer rather than the provider?

    1. Certifying the provider's data centre chain of custody
    2. Executing the provider's backup rotation schedule
    3. Identifying and preserving relevant custodian data in the service Correct answer.
    4. Preserving the provider's internal administrative logs and audit trails

    Explanation

    In eDiscovery the customer, as the party subject to the legal obligation, must identify relevant custodians and data and issue preservation instructions, often using the provider's export and hold tooling. The provider supports this by maintaining platform integrity, preserving data under its control and supplying administrative records when required.

  52. Question 52Risk, Audit, & Compliance

    A company uses a public cloud service to process personal data about its own customers. Under the GDPR, what role does the company generally hold?

    1. A controller, with the cloud provider as processor Correct answer.
    2. A processor, with the cloud provider as controller
    3. A joint controller with the cloud provider in all cases
    4. A supervisory authority delegate

    Explanation

    The customer determines the purposes and means of processing and is therefore the controller, while the cloud provider processes the data on the customer's instructions and acts as a processor. Their relationship is governed by a data processing agreement, and joint controllership arises only in specific circumstances rather than automatically.

  53. Question 53Risk, Audit, & Compliance

    Why should a cloud exit strategy be documented before migrating to a provider?

    1. To satisfy the provider's own ISO/IEC 27001 obligations
    2. To reduce the number of controls the customer must implement
    3. To ensure data and services can be retrieved if the relationship ends Correct answer.
    4. To guarantee that the provider will never raise its prices during the contract term

    Explanation

    An exit strategy defines usable data export formats, transition steps and what happens to data at termination, which reduces lock-in and supports availability and compliance obligations such as return or deletion of personal data. It is a customer-side governance control and does not alter the provider's pricing or certification obligations.

  54. Question 54Risk, Audit, & Compliance

    Which statement accurately describes FedRAMP?

    1. It is a self-certification programme in which each agency attests to its own cloud security posture
    2. It is a European Union regulation for cross-border data transfers
    3. It is a commercial certification limited to financial services providers
    4. It is a government-wide programme for standardizing cloud security assessment and authorization Correct answer.

    Explanation

    FedRAMP provides a standardized, government-wide approach to security assessment, authorization and continuous monitoring of cloud services used by US federal agencies, built on NIST standards and delivered through accredited third-party assessment organizations. It is not a self-certification scheme, and it is distinct from commercial certifications and EU regulations.

  55. Question 55Organization Management

    Which type of governance document establishes management's intent and the mandatory high-level requirements for information security across an organization?

    1. Policy Correct answer.
    2. Guideline
    3. Standard
    4. Procedure

    Explanation

    The document hierarchy places policy at the top: it states management intent and mandatory requirements, and all lower documents must be consistent with it. Standards, procedures, and guidelines exist to implement and explain the policy rather than to set direction.

  56. Question 56Organization Management

    In a data governance model for cloud-hosted information, which role is accountable for classifying a data set and deciding who may access it?

    1. The security operations team
    2. The data owner Correct answer.
    3. The cloud service provider
    4. The data custodian

    Explanation

    The data owner is the business-side role accountable for classification and access decisions, while custodians implement and operate the controls the owner specifies. Providers and operational teams execute technical controls but do not make business decisions about who may see the data.

  57. Question 57Organization Management

    What is the primary objective of an ongoing security awareness program?

    1. To reduce risky employee behavior over time Correct answer.
    2. To document each job role's security duties
    3. To satisfy annual auditor evidence requests
    4. To replace technical controls when budgets are tight

    Explanation

    Awareness programs exist to change behavior, for example improving phishing reporting and data handling, and they work only when delivered continuously rather than as a single annual event. Defining role duties is a job-description or responsibility-matrix activity, and awareness never substitutes for technical controls.

  58. Question 58Organization Management

    A risk assessment shows a low-likelihood exposure whose remediation cost exceeds the potential loss, so management formally documents a decision to live with it and review it annually. Which risk treatment has been applied?

    1. Mitigation
    2. Transference
    3. Avoidance
    4. Acceptance Correct answer.

    Explanation

    Acceptance is a deliberate, documented management decision to retain a risk, typically justified by cost-benefit analysis and paired with periodic review. Avoidance removes the activity, transference shifts financial impact such as through insurance, and mitigation reduces likelihood or impact through controls.

  59. Question 59Organization Management

    Who is normally accountable for approving the organization's security risk appetite and tolerance levels?

    1. Executive management and the board Correct answer.
    2. The security awareness coordinator
    3. Individual system administrators
    4. External penetration testers

    Explanation

    Risk appetite expresses how much risk the organization will take to pursue its objectives, so it must be owned where business strategy is set: executive management with board oversight. Operational and assurance roles apply or test the appetite but do not define it.

  60. Question 60Organization Management

    Which statement correctly distinguishes ISO/IEC 27001 from ISO/IEC 27002?

    1. 27001 sets certifiable requirements; 27002 gives guidance Correct answer.
    2. 27001 is mandatory for cloud providers while 27002 is voluntary
    3. 27001 lists technical controls only; 27002 covers physical controls
    4. 27001 applies to providers while 27002 applies to customers

    Explanation

    ISO/IEC 27001 specifies requirements for an information security management system that can be independently certified, whereas ISO/IEC 27002 is a code of practice offering guidance on selecting and implementing controls. Neither standard is legally mandatory, and both are written to apply to any type and size of organization.

  61. Question 61Organization Management

    What is the primary purpose of the CSA Cloud Controls Matrix?

    1. To provide a template service level agreement
    2. To replace a provider's ISO/IEC 27001 certificate
    3. To benchmark provider pricing against peers
    4. To provide a cloud-specific controls catalogue Correct answer.

    Explanation

    The Cloud Controls Matrix is a cloud-focused catalogue of security controls, organized by domain and mapped to other frameworks such as ISO/IEC 27001, PCI DSS, and NIST, so that provider assessments can be structured and compared. It complements certifications and addresses neither pricing nor contract language.

  62. Question 62Organization Management

    A customer wants structured, consistent answers from a prospective cloud provider about how each control in the Cloud Controls Matrix is implemented. Which artifact should the customer request from the provider?

    1. The CAIQ self-assessment questionnaire Correct answer.
    2. The STAR Level 3 continuous monitoring feed
    3. The CSA Top Threats report
    4. The Cloud Controls Matrix itself

    Explanation

    The Consensus Assessments Initiative Questionnaire is the questionnaire form of the CCM, consisting of control questions a provider answers to describe its security posture, and it is commonly published in the STAR registry. The matrix is the framework being asked about, not the answer format.

  63. Question 63Organization Management

    What does residual risk represent?

    1. The risk left once existing controls are applied Correct answer.
    2. The share of risk transferred to a vendor by contract
    3. The risk a new provider brings before it is assessed
    4. The total risk before any treatment is considered

    Explanation

    Residual risk is the exposure that remains after controls are in place, and it is the figure compared against the risk tolerance before management accepts the risk. The pre-control level is inherent risk, and contractual transference is one of the treatment options rather than a measure of what remains.

  64. Question 64Organization Management

    An assessor ranks likelihood and impact using scales such as low, medium, and high without assigning monetary values. Which approach is being used?

    1. Monte Carlo simulation
    2. Quantitative risk analysis
    3. Qualitative risk analysis Correct answer.
    4. Annualized loss expectancy modeling

    Explanation

    Qualitative analysis uses descriptive rating scales and is useful for rapid prioritization and communication with business stakeholders. Quantitative methods assign numeric or monetary values and support calculations such as annualized loss expectancy or probabilistic simulation.

  65. Question 65Organization Management

    An organization is applying a risk-based approach to third-party assessment and cannot assess every cloud service to the same depth. Which factor should most strongly determine the depth of due diligence?

    1. The provider's pricing relative to competitors
    2. The data classification and service criticality Correct answer.
    3. The number of staff the provider employs
    4. The provider's brand recognition and market share

    Explanation

    Assessment depth should be proportional to the sensitivity of the data involved and the criticality of the business process the service supports, because those drive potential impact. Provider size, brand, and price do not indicate how much harm a failure would cause the customer.

  66. Question 66Organization Management

    Which contract provision best protects a customer when a cloud provider uses downstream subcontractors to process customer data?

    1. Excluding subcontractors from security duties
    2. Free substitution of subcontractors at any time
    3. Annual summary notice after the subcontractor change
    4. Notice of subprocessors with flow-down terms Correct answer.

    Explanation

    Customers need advance visibility of who will handle their data and assurance that the same security and privacy obligations follow the data downstream, which requires notification or approval plus flow-down clauses. Unilateral substitution or after-the-fact summaries remove the customer's ability to object.

  67. Question 67Organization Management

    A security team reports two metrics: the percentage of critical servers with current patches, and the number of confirmed breaches in the last quarter. Which description is accurate?

    1. The patch figure is leading; the incident count is lagging Correct answer.
    2. The patch figure is lagging; the incident count is leading
    3. Both figures are lagging because both are measured after the fact
    4. Both figures are leading indicators of future incidents

    Explanation

    Leading indicators measure the health of preventive controls and predict future outcomes, while lagging indicators measure outcomes that have already occurred. Patch coverage describes the current state of a preventive control, whereas breach counts record events that already happened.

  68. Question 68Organization Management

    A customer wants direct assurance about a provider's controls, but the provider refuses on-site audits and offers only its independent third-party reports. What is the most appropriate governance response?

    1. Accept the refusal because cloud providers never allow on-site audits
    2. Evaluate the attestations against requirements and record the decision Correct answer.
    3. Ask other customers to pressure the provider into allowing audits
    4. Insist on unlimited on-site access as a precondition for signing

    Explanation

    Many providers substitute independent attestations for direct audit rights, so the customer's governance process should determine whether that evidence covers its requirements and formally record acceptance of any gap. Neither blanket acceptance nor an inflexible demand reflects a risk-based decision.

  69. Question 69Organization Management

    Which pairing of a CSA STAR level with its assurance mechanism is correct?

    1. Level 3 – ISO/IEC 27001 certification
    2. Level 3 – continuous CCM monitoring Correct answer.
    3. Level 1 – independent third-party audit
    4. Level 2 – self-assessment questionnaire

    Explanation

    STAR Level 1 is a self-assessment published by the provider, Level 2 requires independent third-party assessment against the CCM (through certification or attestation), and Level 3 adds continuous, automated monitoring of controls. ISO/IEC 27001 certification is one route used at Level 2, not a Level 3 mechanism.

  70. Question 70Organization Management

    Business units may select their own SaaS services, provided each service meets a centrally defined security baseline and is registered in a central inventory. Which governance model does this describe?

    1. Federated governance Correct answer.
    2. Ad hoc governance
    3. Outsourced governance
    4. Centralized governance

    Explanation

    Federated governance balances central standards with local autonomy: the center sets baselines and visibility requirements while business units make service choices within them. Centralized governance would place the selection decision itself with the center, and ad hoc governance would impose no baseline at all.

  71. Question 71Organization Management

    In the three lines of defense model, which group provides independent assurance over the effectiveness of security controls to the board?

    1. The risk management function
    2. Business unit managers
    3. Internal audit Correct answer.
    4. The security operations centre

    Explanation

    The first line is operational management that owns and runs controls, the second line is the risk, compliance, and security functions that set frameworks and monitor, and the third line is internal audit, which reports independently and gives assurance to the board. Independence from management is what qualifies internal audit for the third line.

  72. Question 72Organization Management

    A provider supplies a SOC 2 Type II report, but the system description shows the report covers only its payment processing environment, not the platform the customer uses. What is the most appropriate conclusion?

    1. The report can be relied on whenever the opinion is unqualified
    2. The report fully covers the customer's use because it is the same provider
    3. The report also proves the provider holds ISO/IEC 27001 certification
    4. The report gives no assurance for that service and more evidence is needed Correct answer.

    Explanation

    Attestation reports provide assurance only for the systems and controls described in their scope, so an unqualified opinion about a different environment says nothing about the platform in use. The customer should request evidence covering the actual service or treat the gap as an unassessed risk.

  73. Question 73Identity & Access Management

    What is the primary purpose of identity federation in cloud environments?

    1. To accept authentication from a trusted external identity provider Correct answer.
    2. To replicate user directories into each provider's local database
    3. To encrypt identity data at rest across multiple cloud regions
    4. To convert stored passwords into one-way cryptographic hashes

    Explanation

    Federation establishes a trust relationship in which one party (the identity provider) authenticates a subject and another party (the relying party or service provider) accepts that authentication result. Credentials stay with the identity provider, so users can reach cloud services without a separate account per provider. Encryption, directory replication and password hashing are separate controls that do not constitute federation.

  74. Question 74Identity & Access Management

    OAuth 2.0 is best described as which of the following?

    1. An authentication protocol that asserts a verified user identity to a relying party
    2. A directory synchronization protocol used between identity stores
    3. An encryption standard for protecting data in transit between services
    4. Delegated authorization that issues scoped tokens without sharing credentials Correct answer.

    Explanation

    OAuth 2.0 lets a resource owner authorize a client to access protected resources by issuing an access token with a defined scope, without handing over the owner's credentials. It is an authorization framework and does not by itself assert who the user is; identity assertion is layered on top by OpenID Connect.

  75. Question 75Identity & Access Management

    In SAML 2.0 single sign-on, what does the identity provider deliver to the service provider?

    1. An X.509 certificate revocation list published by the provider
    2. An OAuth refresh token issued by the authorization server
    3. A Kerberos ticket-granting ticket for the user session
    4. A signed assertion containing authentication and attribute statements Correct answer.

    Explanation

    The identity provider digitally signs a SAML assertion that states the authentication event and carries attribute statements about the subject, and returns it in a response to the service provider. The service provider validates the signature and the assertion conditions before establishing a session.

  76. Question 76Identity & Access Management

    Which of the following combinations constitutes true multi-factor authentication?

    1. Two different passwords entered in sequence
    2. A password combined with a memorized security question
    3. A password combined with a one-time code from an authenticator app Correct answer.
    4. A username combined with a numeric PIN

    Explanation

    Multi-factor authentication requires factors drawn from different categories, typically something you know plus something you possess or are. A password combined with a one-time code generated by a registered authenticator app satisfies two distinct categories. Repeating the same category, such as two passwords or a password plus a security question, does not.

  77. Question 77Identity & Access Management

    Which principle requires that each identity receive only the access needed for its intended function, and only for as long as needed?

    1. Zero trust
    2. Separation of duties
    3. Least privilege Correct answer.
    4. Defense in depth

    Explanation

    Least privilege limits entitlements to the minimum necessary for the task and constrains their duration, reducing both the likelihood and the blast radius of misuse. Separation of duties instead splits conflicting tasks among different people, while defense in depth layers controls and zero trust removes implicit trust based on network location.

  78. Question 78Identity & Access Management

    An application running on a cloud virtual machine must read objects from a storage bucket, and the team wants no long-term credentials stored on the instance. Which approach best follows IAM best practice?

    1. Store a service account password in an environment variable on the host
    2. Attach an instance role so the platform issues short-lived credentials Correct answer.
    3. Create a shared user account whose credentials are copied to every instance
    4. Embed a static access key and secret in the application configuration file

    Explanation

    Workload identity mechanisms such as instance roles or service identities let the platform deliver short-lived, automatically rotated credentials that are bound to the workload. This removes static secrets from configuration and images. Embedded keys, stored passwords and shared accounts all create long-lived secrets that are difficult to rotate and attribute.

  79. Question 79Identity & Access Management

    A third-party vendor needs programmatic access to resources in your cloud account. Which approach limits risk most effectively?

    1. Share your own administrator credentials under a signed support agreement
    2. Allowlist the vendor's IP range and issue a long-lived API key
    3. Create a scoped role the vendor assumes with an external ID condition Correct answer.
    4. Create a local user for the vendor and email the credentials to its engineers

    Explanation

    An assumable role scoped to the needed permissions lets the vendor use its own identities while your trust policy constrains who may assume it. Adding an external ID condition mitigates the confused deputy problem in which another customer of the vendor could otherwise reach your resources. Local users, shared administrator credentials and long-lived keys all create unmanaged secrets in someone else's hands.

  80. Question 80Identity & Access Management

    Which standard is designed to automate the exchange of user and group identity data between an identity provider and cloud applications?

    1. OAuth 2.0
    2. Kerberos
    3. SAML 2.0
    4. SCIM Correct answer.

    Explanation

    SCIM defines a REST-based schema and set of operations for creating, updating and deleting identity records, which is what allows automated provisioning and deprovisioning. SAML and OAuth address authentication and delegated authorization at runtime rather than lifecycle management of accounts.

  81. Question 81Identity & Access Management

    A company wants users to sign in to a SaaS application with their corporate identity, and the application must receive verified claims about who the user is and how they authenticated. Which approach fits best?

    1. OAuth 2.0 alone, treating the access token as proof of the user's identity
    2. An LDAP bind forwarded directly from the browser to the SaaS provider
    3. OpenID Connect, which adds an identity layer and ID token to OAuth 2.0 Correct answer.
    4. SAML artifact resolution without a signed assertion

    Explanation

    OpenID Connect extends OAuth 2.0 with an ID token, a signed JWT carrying claims about the authentication event and the subject. That gives the relying application verified identity information rather than only an authorization grant. Using an access token as an identity assertion confuses authorization with authentication and is explicitly discouraged.

  82. Question 82Identity & Access Management

    An organization wants access decisions to reflect device compliance, user risk and location in addition to the credentials presented. Which control best meets this requirement?

    1. Longer session lifetimes to reduce re-authentication prompts
    2. Shared team accounts for staff who work across locations
    3. Static role assignments reviewed once a year by resource owners
    4. Context-aware conditional access policies evaluated at each sign-in Correct answer.

    Explanation

    Conditional access evaluates signals at request time, such as device state, sign-in risk and network location, and can allow, block or step up authentication accordingly. Static periodic reviews and longer sessions do not adapt to the context of an individual request.

  83. Question 83Identity & Access Management

    Which capability set is most characteristic of a privileged access management solution?

    1. Federating consumer social logins into a customer portal
    2. Encrypting data at rest with customer-managed keys
    3. Credential vaulting, just-in-time elevation and admin session monitoring Correct answer.
    4. Self-service password reset for standard users and directory synchronization

    Explanation

    Privileged access management focuses on the accounts that can change critical systems: it vaults their credentials, grants elevation only for an approved window, and records or monitors sessions for accountability. Standard user self-service and social federation address different populations and risks.

  84. Question 84Identity & Access Management

    An employee moves from finance to marketing and later leaves the company. Which practice best addresses the access risk this creates?

    1. Automating entitlement changes from an authoritative identity source Correct answer.
    2. Granting all employees the same baseline access so transfers need no changes
    3. Relying on the user's manager to request removal of unneeded access
    4. Retaining the original account for audit and creating a second account for the new role

    Explanation

    Joiner-mover-leaver processes should be driven from the authoritative HR or directory record so that a role change re-baselines entitlements and a termination revokes them promptly. Manual or duplicated accounts leave finance permissions in place after the move and active credentials after departure.

  85. Question 85Identity & Access Management

    A microservice accepts OAuth 2.0 access tokens from several clients. Which control most directly prevents a token minted for a different service from being accepted?

    1. Storing tokens in browser local storage so they persist across sessions
    2. Increasing token lifetimes so that fewer refreshes are required
    3. Logging every token in plaintext so incidents can be correlated later
    4. Rejecting tokens whose audience claim names a different resource Correct answer.

    Explanation

    The audience claim binds a token to the resource server it was issued for, so validating it ensures a token stolen from or shared by another service cannot be replayed here. Short token lifetimes further shrink the window of misuse, whereas longer lifetimes and plaintext logging increase exposure.

  86. Question 86Identity & Access Management

    A service provider supports SAML single sign-on. Which control best mitigates an attacker replaying a captured assertion?

    1. Requiring signed assertions and matching InResponseTo to a pending request Correct answer.
    2. Extending the assertion validity window to several days for reliability
    3. Disabling signature validation to improve login performance
    4. Accepting assertions over either the HTTP-POST or HTTP-Redirect binding

    Explanation

    Replay protection combines signature validation with binding the assertion to a request the service provider actually issued, typically through InResponseTo and one-time use, and enforcing the NotOnOrAfter condition. A wide validity window or missing signature check lets a captured assertion be reused until it expires.

  87. Question 87Identity & Access Management

    Which statement accurately distinguishes attribute-based access control from role-based access control?

    1. RBAC evaluates environmental attributes such as device posture and time of day at request time
    2. Both models require a shared administrator credential to evaluate policy
    3. ABAC assigns permissions only through static job-function roles
    4. ABAC evaluates subject, resource and context attributes; RBAC uses assigned roles Correct answer.

    Explanation

    RBAC grants permissions by assigning identities to roles that represent job functions, while ABAC writes policies over attributes of the subject, the resource and the environment, allowing finer-grained and context-sensitive decisions. Neither model depends on a shared administrator credential.

  88. Question 88Identity & Access Management

    Which practice best protects a cloud provider's root or global administrator account?

    1. Reserving it for break-glass use with hardware MFA and no access keys Correct answer.
    2. Using it for daily administrative tasks so that permissions stay simple
    3. Disabling MFA on it so that automated scripts can sign in
    4. Storing its credentials in a shared team wiki so they are always available

    Explanation

    The root or global administrator identity bypasses ordinary permission boundaries, so it should be used only when no other path exists, protected by a strong second factor, and stripped of long-lived programmatic credentials. Routine work should be performed through scoped roles or federated identities that are attributable and revocable.

  89. Question 89Security Monitoring

    An organization must be able to produce cloud audit logs on demand for seven years to satisfy a regulator. Which approach best meets this requirement?

    1. Keep logs only in the provider's console, which retains them indefinitely.
    2. Rely on the provider to keep logs for the full regulatory period.
    3. Export logs to immutable, long-term storage with integrity verification. Correct answer.
    4. Delete logs after 90 days and rely on backups taken each quarter.

    Explanation

    Providers retain service logs for defined, often short periods, and the customer remains accountable for meeting regulatory retention obligations. Exporting to immutable, integrity-checked storage keeps records both available and unaltered for the required period.

  90. Question 90Security Monitoring

    In an IaaS deployment, which monitoring activity is primarily the customer's responsibility?

    1. Monitoring the provider's hypervisor patch levels.
    2. Monitoring guest operating system logs and application activity. Correct answer.
    3. Monitoring the physical access controls of the data center.
    4. Monitoring the provider's internal network backbone.

    Explanation

    In IaaS the provider secures the facility, hypervisor, and backbone, while the customer secures the guest OS, middleware, applications, and the logs those layers produce. Monitoring guest OS and application activity therefore falls to the customer.

  91. Question 91Security Monitoring

    What is the primary function of a SIEM in a cloud environment?

    1. Replacing the need for provider-native audit logs.
    2. Enforcing network segmentation between virtual networks.
    3. Correlating events from multiple sources to identify patterns of interest. Correct answer.
    4. Encrypting data at rest across cloud storage buckets.

    Explanation

    A SIEM collects and normalizes events from many sources and applies correlation rules to surface meaningful patterns and incidents. It complements, rather than replaces, provider-native audit logs and does not encrypt data or segment networks.

  92. Question 92Security Monitoring

    Why is consistent time synchronization important for cloud log analysis?

    1. It is required to encrypt log traffic in transit.
    2. It reduces the storage cost of archived log files.
    3. It prevents the provider from accessing customer logs.
    4. It allows events from different systems to be ordered correctly. Correct answer.

    Explanation

    Reconstructing an incident across hosts, services, and regions depends on comparable timestamps. Consistent NTP synchronization prevents misleading event ordering during an investigation.

  93. Question 93Security Monitoring

    Which practice best helps demonstrate that cloud logs have not been altered after collection?

    1. Enabling verbose debug logging across all production systems.
    2. Restricting log access to the security team only.
    3. Hashing log entries and storing the hashes separately. Correct answer.
    4. Compressing logs before sending them to storage.

    Explanation

    Cryptographic hashes computed at collection and stored separately let investigators later prove that entries were not modified. Access restriction and compression are useful controls but do not by themselves demonstrate integrity.

  94. Question 94Security Monitoring

    An enterprise runs workloads in two public clouds and an on-premises data center. What is the most effective way to support cross-environment investigations?

    1. Review each provider's console separately during each incident.
    2. Centralize normalized logs with shared time and identity fields. Correct answer.
    3. Store each environment's logs only in the region where they were created.
    4. Standardize on a single cloud provider to eliminate log variety.

    Explanation

    Investigations spanning environments need a single normalized view with consistent time and identity fields so events can be joined. Reviewing consoles separately slows response and hides cross-cloud attack paths.

  95. Question 95Security Monitoring

    In public cloud platforms, which log source typically records control-plane actions such as creating a virtual machine or changing a firewall rule?

    1. Guest operating system event logs
    2. Provider API activity (audit) logs Correct answer.
    3. Web server access logs
    4. Application performance monitoring traces

    Explanation

    Provider API activity logs record management actions taken through the control plane, including the identity, source address, and target resource. Guest OS, application, and web logs do not capture provider-level configuration changes.

  96. Question 96Security Monitoring

    How does user and entity behavior analytics (UEBA) detect threats in cloud environments?

    1. By enforcing multi-factor authentication at the identity provider.
    2. By scanning storage buckets for known malware signatures.
    3. By modeling normal behavior and flagging unusual activity. Correct answer.
    4. By blocking all traffic that does not match a static allow list.

    Explanation

    UEBA establishes behavioral baselines per user and entity and scores deviations from them, surfacing compromised accounts or insider misuse that static rules miss. It complements rather than replaces preventive controls such as MFA or allow lists.

  97. Question 97Security Monitoring

    A company uses a SaaS application. Which monitoring data is the customer typically able to access directly?

    1. Raw database transaction logs from the provider's backend platform.
    2. Audit logs and user activity reports the vendor exposes. Correct answer.
    3. Host-based intrusion alerts from the SaaS provider's servers.
    4. Hypervisor-level network packet captures of the SaaS infrastructure.

    Explanation

    SaaS customers receive vendor-provided audit and usage records through consoles, APIs, or exports, but not infrastructure-level telemetry from the provider's servers. Visibility is limited to what the vendor exposes contractually and technically.

  98. Question 98Security Monitoring

    Litigation is reasonably anticipated involving a cloud-hosted application. What should the security team do first regarding logs?

    1. Continue normal rotation and rely on provider backups for retrieval.
    2. Move all logs to the cheapest available storage tier immediately.
    3. Suspend deletion of relevant logs and place them under legal hold. Correct answer.
    4. Delete log records containing personal data to reduce exposure.

    Explanation

    Once litigation is reasonably anticipated, relevant records must be preserved, which means halting routine deletion and applying a legal hold. Continuing rotation or deleting data risks spoliation of evidence.

  99. Question 99Security Monitoring

    A team needs to inspect traffic between virtual machines inside a cloud VPC for intrusion detection. Which capability is most appropriate?

    1. Provider traffic mirroring or packet capture features. Correct answer.
    2. Installing a hardware tap on the provider's backbone link.
    3. Enabling verbose logging on the hypervisor management plane.
    4. Spanning the physical switch ports of the provider's network.

    Explanation

    Cloud providers expose traffic mirroring or packet capture services that copy instance traffic to an inspection target such as an IDS appliance. Physical taps and switch spanning are not available to tenants in public cloud.

  100. Question 100Security Monitoring

    A SIEM generates hundreds of identical low-severity alerts each day from a known authorized vulnerability scanner. What is the best response?

    1. Disable the alerting rule entirely and stop collecting the source logs.
    2. Raise the severity of the rule so analysts notice it more.
    3. Ignore all alerts from the SIEM for the next quarter.
    4. Tune the rule with exclusions or thresholds for the known scanner. Correct answer.

    Explanation

    Alert quality depends on tuning rules to the environment, using exclusions, suppressions, or thresholds so known benign activity does not drown out real signals. Blanket disabling or ignoring alerts removes detection coverage.

  101. Question 101Security Monitoring

    Which statement best distinguishes cloud monitoring metrics from logs?

    1. Metrics are only produced by applications, never by cloud services.
    2. Metrics contain full event text used for forensic reconstruction.
    3. Metrics are always retained longer than logs by providers.
    4. Metrics are numeric time series used for dashboards and alarms. Correct answer.

    Explanation

    Metrics are aggregated numeric time series suited to dashboards, capacity planning, and threshold alarms, while logs are detailed event records used for investigation. The two are complementary and are usually correlated together.

  102. Question 102Security Monitoring

    To make cloud audit logs defensible as evidence, which combination is most appropriate?

    1. Relying on provider-side retention alone, since providers are audited.
    2. Storing logs only on analyst laptops and relying on file timestamps.
    3. Encrypting logs with a key held by the same administrator who can delete them.
    4. Append-only, hash-verified storage with documented access controls. Correct answer.

    Explanation

    Defensible evidence requires an unbroken chain: continuous forwarding, restricted append-only storage, hash verification, and documented access. Encryption alone, or provider retention alone, does not prevent tampering or deletion by privileged insiders.

  103. Question 103Security Monitoring

    Which monitoring approach best addresses ephemeral serverless functions that cannot run a traditional host agent?

    1. Instrumenting code and ingesting platform logs and invocation metrics. Correct answer.
    2. Relying on network packet capture at the function's virtual NIC.
    3. Installing a kernel-level agent inside each function runtime.
    4. Capturing full memory dumps from each function instance for later review.

    Explanation

    Serverless functions are short-lived and fully managed, so monitoring relies on code instrumentation plus platform logs, traces, and invocation metrics. Host agents, memory dumps, and NIC-level capture are impractical or unavailable to tenants.

  104. Question 104Security Monitoring

    Why should monitoring cover data-plane events in addition to control-plane API calls in object storage?

    1. Control-plane audit logs never contain identity or session information.
    2. Data-plane events are always much lower in volume than control-plane events.
    3. Bulk object reads can indicate exfiltration without configuration changes. Correct answer.
    4. Data-plane events are the only ones providers audit by default and retain.

    Explanation

    Configuration changes appear in control-plane logs, but mass object reads or downloads occur in the data plane and can signal exfiltration even when no settings change. Comprehensive monitoring therefore covers both planes.

  105. Question 105Infrastructure & Networking

    In a traditional data center, which capability does a VLAN primarily provide?

    1. Encryption of all traffic between hosts that share the same subnet
    2. Routing of packets between different IP subnets
    3. Logical separation of Layer 2 broadcast domains Correct answer.
    4. Authentication of users before they reach a server

    Explanation

    VLANs tag frames at Layer 2 so a single physical switch fabric can carry multiple logically separate broadcast domains. Hosts in different VLANs cannot reach each other without a Layer 3 device, which makes VLANs a segmentation control rather than an encryption or authentication control.

  106. Question 106Infrastructure & Networking

    What is the key difference between a cloud security group and a network ACL?

    1. Security groups are stateful; network ACLs are stateless Correct answer.
    2. Security groups operate at the subnet level; network ACLs operate at the instance level
    3. Security groups apply only to outbound traffic; network ACLs apply only to inbound traffic
    4. Network ACLs can inspect application payloads; security groups cannot

    Explanation

    Security groups track connection state, so a permitted flow is automatically allowed for return traffic. Network ACLs evaluate each packet independently in each direction and therefore require explicit rules for both inbound and outbound traffic.

  107. Question 107Infrastructure & Networking

    What is the primary role of a scrubbing center in DDoS defence?

    1. To filter malicious traffic before it reaches the protected network Correct answer.
    2. To store forensic packet captures for later legal action
    3. To distribute authoritative DNS answers across many locations
    4. To issue and manage TLS certificates for public web endpoints

    Explanation

    Scrubbing centers are inline filtering points: traffic is diverted to them, attack traffic is dropped, and clean traffic is forwarded to the origin. They are sized and distributed so that volumetric floods are absorbed away from the protected asset.

  108. Question 108Infrastructure & Networking

    Which statement accurately distinguishes a Type 1 from a Type 2 hypervisor?

    1. A Type 1 runs directly on hardware; a Type 2 runs on a host OS Correct answer.
    2. A Type 1 hypervisor is always hosted in the public cloud; a Type 2 is always on-premises
    3. A Type 1 hypervisor only supports containers; a Type 2 only supports virtual machines
    4. A Type 1 hypervisor requires a host operating system; a Type 2 runs on bare metal

    Explanation

    The distinction is architectural: a Type 1 hypervisor sits directly on the physical hardware, while a Type 2 hypervisor is an application running on top of a general-purpose operating system. The narrower attack surface of Type 1 designs is one reason they dominate server virtualisation.

  109. Question 109Infrastructure & Networking

    An organisation wants to limit lateral movement between workloads contained in a single subnet. Which approach best achieves this?

    1. Require multi-factor authentication for all administrative logins
    2. Enforce host-level policy between individual workloads Correct answer.
    3. Deploy a perimeter firewall at the internet edge of the data center
    4. Segment the subnet into smaller VLANs based on physical rack location

    Explanation

    Microsegmentation places policy at the workload level so that traffic between any two hosts is explicitly permitted or denied, regardless of subnet boundaries. This directly constrains east-west movement inside a segment, which perimeter controls cannot see.

  110. Question 110Infrastructure & Networking

    An enterprise connects 40 VPCs in the same region to a shared services VPC. Which design scales with the fewest connections to manage?

    1. A single site-to-site VPN tunnel shared by every VPC
    2. One internet gateway attached to each individual VPC
    3. A hub-and-spoke topology built on a transit gateway Correct answer.
    4. A full mesh of VPC peering connections between every pair of VPCs

    Explanation

    A transit gateway acts as a regional routing hub, so each spoke VPC needs only one attachment instead of a peering link to every other VPC. This keeps the connection count linear rather than quadratic as the number of VPCs grows.

  111. Question 111Infrastructure & Networking

    In a software-defined perimeter, what does single packet authorization achieve?

    1. It replaces the need for any authentication of the connecting device
    2. It load-balances inbound connections across a pool of hidden servers
    3. It encrypts every packet with a per-session key negotiated by the gateway
    4. It keeps ports closed until a client proves its identity Correct answer.

    Explanation

    A software-defined perimeter uses a default-deny model in which the protected service does not respond to unauthenticated probes, so the port appears closed. A client sends a signed, validated authorization packet first, and only then is a connection path opened.

  112. Question 112Infrastructure & Networking

    A security team observes many long, randomly named TXT queries to a single external domain from one internal server. Which activity does this most likely indicate?

    1. A zone transfer of the internal DNS records
    2. A cache poisoning attempt against the internal resolver
    3. A reflection attack using spoofed source addresses
    4. Data exfiltration through DNS tunnelling Correct answer.

    Explanation

    DNS tunnelling encodes stolen data in the labels of queries and receives instructions in responses, which is why high volumes of long, high-entropy names against one domain are a classic indicator. TXT records are frequently abused because they can carry larger payloads per answer.

  113. Question 113Infrastructure & Networking

    A company wants visibility into which sanctioned SaaS applications employees use, without routing traffic through an on-premises proxy. Which CASB deployment mode fits?

    1. Log collection from the corporate VPN concentrator only
    2. Forward proxy deployed inline with TLS interception on every endpoint
    3. Reverse proxy placed in front of internal data center applications
    4. API-based access to the SaaS provider's logs Correct answer.

    Explanation

    API-based CASBs connect directly to the sanctioned provider's interfaces and read activity data out-of-band, so no user traffic is redirected. This gives discovery and data-level visibility for sanctioned services, though it cannot see unsanctioned shadow IT.

  114. Question 114Infrastructure & Networking

    Which mechanism allows a network operator to cryptographically attest which autonomous system is authorised to originate a given IP prefix?

    1. Unicast reverse path forwarding on the edge router
    2. BGP prefix-limit thresholds on each eBGP session
    3. RPKI route origin authorisations Correct answer.
    4. DNSSEC validation at the recursive resolver

    Explanation

    RPKI publishes signed route origin authorisations that bind a prefix to the autonomous system permitted to announce it. Routers performing route origin validation can then drop or down-rank announcements that contradict the signed record, mitigating hijacks.

  115. Question 115Infrastructure & Networking

    What is the defining characteristic of software-defined networking?

    1. Network functions are delivered only as virtual machines on a hypervisor
    2. Traffic is always encrypted between every pair of endpoints
    3. The control plane is decoupled from the data plane and programmed centrally Correct answer.
    4. All packet forwarding decisions are made by a distributed routing protocol on each device

    Explanation

    SDN separates the logic that decides where traffic goes from the devices that actually forward it, exposing a central programmable controller. That separation allows network-wide policy to be expressed in software rather than configured device by device.

  116. Question 116Infrastructure & Networking

    In a VXLAN overlay, which element identifies the isolated Layer 2 segment to which a frame belongs?

    1. The 24-bit VXLAN network identifier in the header Correct answer.
    2. The 802.1Q tag on the physical uplink trunk
    3. The outer UDP source port chosen by the sending VTEP
    4. The outer destination IP address of the tunnel endpoints

    Explanation

    VXLAN encapsulates the original frame and carries a 24-bit network identifier that scopes the inner Layer 2 domain. Because that identifier rides inside the tunnel header, the underlay IP fabric can carry many isolated segments, far more than 802.1Q VLAN IDs allow.

  117. Question 117Infrastructure & Networking

    What is the primary purpose of a web application firewall?

    1. Caching static content to reduce origin server load
    2. Inspecting HTTP requests and blocking application-layer attacks Correct answer.
    3. Filtering traffic by IP address and port at the network perimeter
    4. Terminating TLS sessions for backend web servers

    Explanation

    A web application firewall parses HTTP and HTTPS conversations and applies rules against injection, cross-site scripting and similar Layer 7 attacks. Network-layer firewalls cannot make these judgements because they do not understand application semantics.

  118. Question 118Infrastructure & Networking

    Which storage model presents block-level volumes over a dedicated network to a single host?

    1. A storage area network Correct answer.
    2. A distributed file system replicated across regions
    3. Object storage accessed through an HTTP API
    4. A network-attached file share using SMB or NFS

    Explanation

    A storage area network delivers raw block devices over a dedicated fabric such as Fibre Channel or iSCSI, and the host's own file system formats them. This differs from file shares, where the storage device owns the file system and many clients mount it.

  119. Question 119Infrastructure & Networking

    A team needs to investigate unusual traffic patterns between instances but must not capture payload contents. Which source is most appropriate?

    1. Application debug logs written by each service
    2. A TLS-terminating proxy in front of the workload
    3. VPC flow logs recording IP metadata Correct answer.
    4. Full packet capture on each hypervisor host

    Explanation

    Flow logs record connection metadata such as source, destination, port, protocol and byte counts, which reveals traffic patterns without storing payload. This satisfies the investigation need while respecting the prohibition on content capture.

  120. Question 120Infrastructure & Networking

    Which connectivity option keeps data center traffic off the public internet between an on-premises site and a cloud provider?

    1. A dedicated private interconnect with the provider Correct answer.
    2. A site-to-site IPsec VPN over broadband
    3. A remote-access SSL VPN for individual users
    4. A cloud NAT gateway with static egress addresses

    Explanation

    A dedicated interconnect provides a private physical or provider-provisioned path between the enterprise edge and the cloud network, so packets never traverse public internet infrastructure. This also typically delivers more predictable latency than internet-based options.

  121. Question 121Cloud Workload Security

    Which best describes a cloud workload for security purposes?

    1. A network segment that connects cloud virtual machines to on-premises data center systems
    2. Any managed storage service holding data that is accessed by applications running in the cloud
    3. An application or service executing on cloud compute, with its code and configuration Correct answer.
    4. The identity and access management policy attached to a cloud resource

    Explanation

    A workload is the running unit of compute - a VM, container, function, or application - together with its code, configuration, and the data it processes. Security controls are applied to the workload's lifecycle: build, deploy, run, and retire. Storage services, network segments, and IAM policies are supporting resources that the workload consumes rather than the workload itself.

  122. Question 122Cloud Workload Security

    Containers on a shared host are isolated primarily through which mechanism?

    1. A hypervisor that gives each container its own virtual hardware
    2. Separate physical network interface cards assigned to each container
    3. Per-container firmware that enforces boot-time integrity measurements
    4. Linux kernel namespaces and control groups that partition host resources Correct answer.

    Explanation

    Containers share the host kernel, so isolation comes from kernel features: namespaces limit what a process can see, and control groups limit what it can consume. Because the kernel is shared, a kernel vulnerability or excessive privileges can weaken isolation. Hypervisor-level separation applies to virtual machines, not standard containers.

  123. Question 123Cloud Workload Security

    In a function-as-a-service offering, which security task remains the customer's responsibility?

    1. Maintaining the physical security of the data center housing the servers
    2. Securing the function code, its dependencies and the permissions granted to it Correct answer.
    3. Updating the language runtime that the provider uses to execute functions
    4. Patching the operating system of the hosts that execute functions

    Explanation

    Serverless shifts host, runtime, and facility management to the provider, but the customer still owns the function code, its third-party libraries, its configuration, and the identity and permissions under which it executes. Over-permissive execution roles and vulnerable dependencies are the dominant risks in FaaS. Operating system, runtime, and physical controls belong to the provider.

  124. Question 124Cloud Workload Security

    Which control best verifies that a container image came from an approved build pipeline and was not altered before deployment?

    1. Scanning the image for known vulnerabilities with a scanner before deployment
    2. Configuring the running container to use a read-only root file system
    3. Signing images and validating the signature at deployment time Correct answer.
    4. Restricting access to the registry using role-based access control policies

    Explanation

    Image signing creates a cryptographic link between the image digest and a trusted builder identity; verification at admission or pull time rejects unsigned or tampered images. Scanning tells you about known flaws, and RBAC limits who can push or pull, but neither proves provenance or integrity of the artifact itself. Signing also anchors the chain of custody for the deployment pipeline.

  125. Question 125Cloud Workload Security

    What is the primary purpose of a CIS Benchmark?

    1. To measure the latency and availability of cloud workloads against an SLA
    2. To certify that a cloud provider's data centers meet physical security requirements
    3. To provide consensus hardening guidance for a platform that can be audited Correct answer.
    4. To define the legal data residency obligations of a cloud service provider

    Explanation

    CIS Benchmarks are consensus-developed, prescriptive configuration baselines for operating systems, services, and cloud platforms. They are written so that automated tools can assess a system against each recommendation and report pass/fail. They are configuration guidance, not certification schemes, legal instruments, or performance metrics.

  126. Question 126Cloud Workload Security

    An operations team rebuilds server images from a hardened template on every deployment instead of patching running instances. What is the main security benefit?

    1. It allows the team to stop maintaining a vulnerability management process
    2. It eliminates configuration drift that accumulates on long-lived hosts Correct answer.
    3. It removes the need to scan images for vulnerabilities before release
    4. It guarantees that workloads are unreachable from the public internet

    Explanation

    Immutable infrastructure replaces hosts rather than modifying them, so every instance matches the known-good template and undocumented changes cannot silently accumulate. The image pipeline still requires scanning, and vulnerability management still applies to the template and its dependencies. Immutability addresses drift and reproducibility, not network exposure.

  127. Question 127Cloud Workload Security

    A container image in a public registry contains a hardcoded API key. Which remediation best addresses the root cause?

    1. Revoke the key and have the workload fetch credentials at runtime from a secrets manager Correct answer.
    2. Encrypt the individual image layers so the key cannot be read by anyone who pulls the image
    3. Restrict who can pull the image so that only production clusters can deploy it into service
    4. Rotate the exposed key and move the image into a private registry with restricted pull access

    Explanation

    Credentials baked into an image persist in every layer and copy and cannot be reliably removed after the fact, so the key must be revoked and removed from the build entirely. Runtime retrieval from a secrets manager, ideally tied to a workload identity, keeps credentials short-lived and out of the artifact. Registry restrictions reduce exposure but do not fix the credential being embedded.

  128. Question 128Cloud Workload Security

    A Kubernetes pod is configured with privileged: true. What is the most significant security consequence?

    1. The pod's container images bypass vulnerability scanning at admission
    2. The container gains host device access and kernel capabilities, weakening isolation Correct answer.
    3. The pod's logs are written to the node's file system instead of the log aggregator
    4. The pod can only be scheduled on nodes that have no network policy applied

    Explanation

    A privileged container receives all Linux capabilities, access to host devices, and relaxed security controls such as mandatory access control profiles, which effectively removes the container boundary. Compromise of such a pod frequently leads to node compromise. Image scanning, log routing, and scheduling constraints are unaffected by the privileged flag.

  129. Question 129Cloud Workload Security

    Which approach most reliably keeps a large fleet of long-lived virtual machines current with security patches?

    1. Deploying patches only after a vulnerability is confirmed to be exploited in the wild
    2. Disabling automatic updates so that unexpected reboots never occur during business hours
    3. Automated patching with defined maintenance windows and continuous compliance checks Correct answer.
    4. Relying on the cloud provider, which patches guest operating systems by default

    Explanation

    For infrastructure-as-a-service, the customer owns guest patching, so the reliable pattern is automation with controlled maintenance windows plus ongoing verification that each host is actually compliant. Waiting for exploitation or disabling updates leaves a known window of exposure. Providers do not patch guest operating systems in IaaS by default.

  130. Question 130Cloud Workload Security

    UEFI Secure Boot contributes to workload integrity mainly by doing what?

    1. Encrypting the disk that holds the operating system image and its configuration
    2. Refusing to load boot components whose signatures do not match trusted keys Correct answer.
    3. Verifying that the workload's application code has no known vulnerabilities
    4. Recording boot measurements in a remote log for later forensic analysis

    Explanation

    Secure Boot validates the signature of each boot component against keys stored in firmware, blocking bootkits and tampered bootloaders. Measurement and logging for attestation are the role of measured boot with a TPM, which is complementary. Full-disk encryption and application vulnerability scanning are unrelated controls.

  131. Question 131Cloud Workload Security

    Two tiers of the same application communicate over the internal network. Which control most directly limits lateral movement between them?

    1. Microsegmentation policies that allow only the specific flows each tier requires Correct answer.
    2. Multi-factor authentication enforced for administrators who log on to the hosts
    3. Web application firewall rules applied at the internet-facing load balancer
    4. Host-based antivirus scanning of files written to each workload's disk

    Explanation

    Microsegmentation applies least-privilege allow-lists to workload-to-workload traffic, so a compromised tier cannot freely reach peers it has no legitimate reason to contact. A web application firewall inspects north-south HTTP traffic, and host antivirus and administrative MFA do not restrict east-west flows. The control must be scoped to the actual application flows to be effective.

  132. Question 132Cloud Workload Security

    How does a sandboxed container runtime such as gVisor reduce risk compared with a standard container runtime?

    1. It interposes a user-space kernel between the container and the host kernel Correct answer.
    2. It encrypts all container-to-container traffic on the node automatically
    3. It scans every container image continuously against a vulnerability feed
    4. It stores container file systems in memory so data cannot persist on disk

    Explanation

    Sandboxed runtimes intercept the container's system calls and service them in a user-space kernel or lightweight virtual machine, shrinking the host kernel attack surface available to a compromised workload. This strengthens isolation without requiring each workload to be a full virtual machine. Traffic encryption, image scanning, and in-memory file systems are separate concerns.

  133. Question 133Cloud Workload Security

    Which capability distinguishes a cloud workload protection platform from a cloud security posture management tool?

    1. Mapping compliance evidence to regulatory frameworks such as PCI DSS
    2. Identifying publicly readable storage buckets and overly permissive role assignments
    3. Detecting runtime threats and behavioural anomalies on running workloads Correct answer.
    4. Benchmarking cloud accounts against CIS Foundations controls on a schedule

    Explanation

    Workload protection focuses on the running compute unit: process behaviour, file integrity, network activity, and malware or exploit detection in real time. Posture management evaluates the configuration of cloud resources and identities against baselines and compliance frameworks. The two are complementary but answer different questions: what is running versus how it is configured.

  134. Question 134Cloud Workload Security

    An organization moves a sensitive data-processing workload into a hardware trusted execution environment. Which statement about the protection gained is accurate?

    1. In-use data is hidden from the host, but flaws inside the enclave's own code remain exploitable Correct answer.
    2. Attestation guarantees that the workload's third-party libraries contain no known vulnerabilities
    3. The enclave removes the need for encryption of the data at rest and in transit around it
    4. The enclave prevents memory-safety bugs in the workload's code from being exploited at runtime

    Explanation

    Confidential computing protects data in use from the host, hypervisor, and cloud operator by isolating memory in a hardware enclave, and attestation lets a relying party verify what code is running. It does not make that code correct or patched, so application and library vulnerabilities inside the enclave still matter. Encryption at rest and in transit remains necessary because the enclave only covers the in-use state.

  135. Question 135Cloud Workload Security

    An admission controller is meant to admit only signed images, yet unsigned images are still being deployed. Which weakness most likely explains this?

    1. The policy is enforced by image tag, and a mutable tag was re-pointed after signing Correct answer.
    2. The registry's vulnerability scanner has not finished scanning the newest images
    3. The cluster's nodes are running a container runtime that does not support signing
    4. The signing key was stored in a hardware security module rather than in a file

    Explanation

    Signatures bind to an immutable image digest, so a control that checks a tag such as latest or v1 can be satisfied by a signed image and then bypassed by pushing different content to the same tag. Enforcement must resolve and verify the digest. Scanning status, runtime signing support, and key storage in an HSM do not create this bypass.

  136. Question 136Cloud Workload Security

    An attacker with code execution inside a container escalates to root on the underlying node. Which configuration most likely enabled this?

    1. The container image was built from a distroless base with no shell installed
    2. The container mounted the host's container runtime socket from the node Correct answer.
    3. The container was started with a read-only root file system and dropped capabilities
    4. The node used a container-optimized operating system with a minimal package set

    Explanation

    Mounting the runtime socket gives a container control over the container runtime, which can be used to launch a privileged container or otherwise take over the host. This is a host-exposure misconfiguration rather than a kernel exploit. Read-only file systems, minimized host images, and distroless bases all reduce attack surface and would not provide host takeover.

  137. Question 137Data Security

    An organization must protect data moving between its on-premises systems and a cloud object store. Which control directly addresses this requirement?

    1. Server-side encryption of the bucket
    2. Object lifecycle rules
    3. TLS for the connection Correct answer.
    4. Bucket versioning

    Explanation

    Encryption in transit protects data as it moves across networks. TLS provides that protection for connections between systems. Server-side encryption protects data at rest after it reaches the bucket.

  138. Question 138Data Security

    Before applying controls to cloud data, what is the primary purpose of a data classification scheme?

    1. To reduce the recurring cost of archived objects
    2. To assign each dataset to a single cloud region
    3. To match controls to data sensitivity Correct answer.
    4. To eliminate encryption for low-value datasets

    Explanation

    Classification assigns sensitivity labels so that encryption, access, retention, and monitoring controls can be chosen proportionately. It is a prerequisite for risk-based control selection.

  139. Question 139Data Security

    Which statement best describes encryption at rest in a cloud environment?

    1. It protects data while it is being transmitted between regions
    2. It renders stored data unreadable without the appropriate key Correct answer.
    3. It replaces the need for access control on stored objects
    4. It automatically classifies data before storage

    Explanation

    Encryption at rest converts stored data into ciphertext so that reading the underlying media without the key does not reveal plaintext. It complements, but does not replace, access control and monitoring.

  140. Question 140Data Security

    A cloud key management service (KMS) is primarily used to do what?

    1. Scan object storage for sensitive data patterns
    2. Replace TLS for service-to-service traffic
    3. Store application secrets in source code repositories
    4. Centrally manage keys and their policies Correct answer.

    Explanation

    A KMS centralizes key creation, storage, rotation, and authorization policies, and typically exposes cryptographic operations through APIs. Applications use it instead of embedding keys locally.

  141. Question 141Data Security

    Which control is designed to detect and block sensitive data such as credit card numbers from leaving a cloud environment?

    1. Data loss prevention Correct answer.
    2. Load balancing
    3. Object versioning
    4. Key rotation

    Explanation

    DLP inspects content and context to identify sensitive data and can alert or block its movement. It is specifically aimed at preventing unauthorized exfiltration.

  142. Question 142Data Security

    An application encrypts each object with a unique data key and then encrypts that data key with a master key held in a KMS. What is this pattern called?

    1. Homomorphic encryption
    2. Envelope encryption Correct answer.
    3. Format-preserving encryption
    4. Tokenization

    Explanation

    Envelope encryption uses a data encryption key per object or dataset and wraps that key with a key encryption key held in a KMS. This limits how much data is exposed if a data key is compromised and keeps bulk data away from the central key service.

  143. Question 143Data Security

    Which statement distinguishes bring your own key (BYOK) from hold your own key (HYOK)?

    1. BYOK keeps all key material outside the cloud provider at all times
    2. HYOK requires the provider to generate and store the root key
    3. HYOK keeps key material outside the provider's key service Correct answer.
    4. BYOK and HYOK both require the provider to escrow the root key

    Explanation

    With BYOK, the customer supplies key material but the provider's key service stores and uses it. With HYOK, the key remains in customer-controlled infrastructure and the provider calls out to it or receives wrapped keys.

  144. Question 144Data Security

    A cloud customer needs to guarantee that encrypted backups are unrecoverable when a contract ends. Which approach most directly achieves this?

    1. Deleting the backup bucket and waiting for provider garbage collection
    2. Destroying the customer-controlled keys used to encrypt the backups Correct answer.
    3. Enabling versioning on the backup bucket
    4. Rotating the master key every 90 days

    Explanation

    Cryptographic erasure relies on destroying the keys so ciphertext becomes permanently unreadable, which is practical when backups are distributed or provider-managed. Deleting objects alone may not remove all copies.

  145. Question 145Data Security

    How does tokenization differ from encryption when protecting sensitive data?

    1. Tokens can be reversed by anyone who knows the algorithm
    2. Tokenization protects data only while it is in transit
    3. Tokenization always uses a key per data record
    4. Tokens are non-sensitive substitutes mapped through a vault Correct answer.

    Explanation

    Tokenization replaces sensitive values with tokens that have no mathematical relationship to the original; a separate token vault maps them back. Encryption, by contrast, is reversible with a key.

  146. Question 146Data Security

    When cloud storage media are retired, what is the most reliable way for a customer to prevent recovery of its data?

    1. Rely on the provider to overwrite the physical disk once
    2. Delete the object and empty the recycle bin
    3. Destroy the keys used to encrypt the data Correct answer.
    4. Rename the objects before deletion

    Explanation

    Destroying the encryption keys makes the ciphertext permanently unreadable, which is a reliable customer-controlled method when physical media are not accessible. Provider sanitization should be verified contractually.

  147. Question 147Data Security

    A security team wants to discover unsanctioned SaaS use and enforce data policies for cloud traffic without proxying all traffic. Which CASB deployment mode fits best?

    1. API-based Correct answer.
    2. Network tap only
    3. Forward proxy
    4. Inline reverse proxy

    Explanation

    API-based CASB connects directly to SaaS provider APIs to discover sanctioned and unsanctioned apps, classify data, and apply DLP without inline traffic redirection. Inline proxy modes inspect traffic in the path.

  148. Question 148Data Security

    Which technology is intended to protect data while it is being processed in the cloud?

    1. TLS termination at the load balancer
    2. Database column masking
    3. Server-side encryption with provider-managed keys
    4. Confidential computing using hardware enclaves Correct answer.

    Explanation

    Confidential computing uses hardware trusted execution environments to isolate data during processing, so the host and provider cannot read plaintext. It complements encryption at rest and in transit.

  149. Question 149Data Security

    Where should an application running in the cloud store database credentials so they are not exposed in code or images?

    1. In a public container registry variable
    2. In the application's source repository
    3. In a shared spreadsheet for operators
    4. In a managed secrets vault at runtime Correct answer.

    Explanation

    A managed secrets vault stores credentials outside code and images and injects them at runtime with access controls and auditing. This reduces exposure in repositories and build artifacts.

  150. Question 150Data Security

    Which statement best captures the difference between data residency and data sovereignty?

    1. Residency is where data is stored; sovereignty is which laws apply Correct answer.
    2. Residency applies only to backups; sovereignty applies only to primary data
    3. Residency concerns encryption strength; sovereignty concerns key length
    4. Residency and sovereignty are synonyms in cloud contracts

    Explanation

    Residency is about the geographic location where data is stored. Sovereignty is about which legal jurisdiction and laws govern the data, regardless of storage location.

  151. Question 151Data Security

    When a KMS master key that wraps many data keys is rotated, what is typically required to keep existing ciphertext readable?

    1. Delete all old data keys immediately
    2. Re-wrap data keys with the new master key Correct answer.
    3. Re-encrypt every object with a new data key
    4. Disable envelope encryption for old objects

    Explanation

    Rotating a master key that only wraps data keys does not require re-encrypting the data; the existing data keys are re-wrapped under the new master key. Re-encryption of data is needed when a data key itself is compromised.

  152. Question 152Data Security

    An object is split into shards with erasure coding and each shard is encrypted with a different key. What is a key data-security consequence?

    1. Compromise of one shard's key exposes only that shard, not the whole object Correct answer.
    2. Loss of any single shard makes the object unreadable
    3. Erasure coding removes the need for key management
    4. The object can be reconstructed only if all shards are decrypted

    Explanation

    With erasure coding, the object can be rebuilt from a subset of shards, and separate per-shard keys mean one key compromise exposes only that shard. Full object recovery would require enough shards and their keys.

  153. Question 153Application Security

    In a secure software development lifecycle, at which point is a design flaw typically least expensive to correct?

    1. During deployment
    2. During integration testing
    3. In requirements and design Correct answer.
    4. After release into production

    Explanation

    Security defects are cheapest to fix when they are caught in requirements and design, because changes are made to specifications and models rather than to working code. Remediation cost rises sharply as the artifact moves through testing, deployment, and production.

  154. Question 154Application Security

    Which STRIDE threat category addresses the ability of a user to deny performing an action?

    1. Tampering
    2. Information disclosure
    3. Elevation of privilege
    4. Repudiation Correct answer.

    Explanation

    Repudiation threats involve actors denying that they performed an action without sufficient evidence to prove otherwise. Tampering covers unauthorized modification, information disclosure covers exposure of data, and elevation of privilege covers gaining unauthorized rights.

  155. Question 155Application Security

    Which control most directly prevents untrusted data from being interpreted as executable script in a web page?

    1. Context-aware output encoding Correct answer.
    2. Rate limiting on the login endpoint
    3. Parameterized database queries
    4. TLS encryption of the session

    Explanation

    Context-aware output encoding converts untrusted data into a safe representation for the specific HTML, attribute, or script context where it is rendered. This ensures the browser treats the data as content rather than executable code.

  156. Question 156Application Security

    Which specification is designed primarily for delegated authorization rather than user authentication?

    1. OAuth 2.0 Correct answer.
    2. OpenID Connect
    3. Kerberos
    4. SAML 2.0

    Explanation

    OAuth 2.0 is an authorization framework that lets a client obtain limited access to a resource on behalf of a user without sharing credentials. OpenID Connect adds an authentication layer on top of OAuth 2.0, while SAML and Kerberos are authentication protocols.

  157. Question 157Application Security

    Where should application secrets such as API keys be stored for a cloud-native delivery pipeline?

    1. In a managed secrets vault Correct answer.
    2. In build log output
    3. In the source code repository
    4. In container image layers

    Explanation

    A managed secrets vault centralizes storage, enforces access control, and supports rotation and auditing of credentials. Secret material should never be embedded in artifacts or logs where it can be widely read or retained.

  158. Question 158Application Security

    Which testing approach analyzes source code without executing the application?

    1. Static application security testing Correct answer.
    2. Dynamic application security testing
    3. Interactive application security testing
    4. Software composition analysis

    Explanation

    Static application security testing examines source, bytecode, or binaries for insecure patterns without running the program. Dynamic testing exercises a running application, interactive testing instruments a live runtime, and composition analysis inventories third-party dependencies.

  159. Question 159Application Security

    A login form concatenates user input directly into a SQL statement. Which remediation removes the vulnerability?

    1. Restricting input to 20 characters
    2. Escaping single quotes in the input
    3. Using parameterized queries Correct answer.
    4. Hiding database errors from users

    Explanation

    Parameterized queries separate SQL code from data, so user input cannot alter the statement's structure. Escaping and length limits are fragile filters, and error suppression only hides symptoms of a successful injection.

  160. Question 160Application Security

    Which control best mitigates server-side request forgery in a cloud application that fetches user-supplied URLs?

    1. Requiring TLS for every outbound request
    2. Enforcing a destination allowlist Correct answer.
    3. Disabling redirects in the HTTP client
    4. Validating URLs with a regex

    Explanation

    An allowlist limits outbound fetches to known, approved destinations and blocks access to internal metadata services and private address ranges. Regex validation is easily bypassed, and TLS or redirect controls do not prevent requests to attacker-chosen internal hosts.

  161. Question 161Application Security

    In OpenID Connect, which token carries signed claims about the authenticated user's identity?

    1. Access token
    2. ID token Correct answer.
    3. Refresh token
    4. Session cookie

    Explanation

    The ID token is a signed JSON Web Token containing claims such as the subject identifier and authentication context. Access tokens authorize API calls, refresh tokens obtain new tokens, and session cookies are application-specific state.

  162. Question 162Application Security

    Which practice best reduces the risk of a compromised build pipeline injecting malicious code into artifacts?

    1. Granting pipeline service accounts broad admin rights
    2. Running builds on developer workstations
    3. Storing build credentials in environment variables
    4. Signing and verifying build artifacts Correct answer.

    Explanation

    Signing artifacts and verifying those signatures before deployment provides integrity and provenance, so tampered artifacts are rejected. Environment variables are not a strong boundary, and broad pipeline permissions increase the impact of a compromise.

  163. Question 163Application Security

    Which mechanism is most effective at preventing cross-site request forgery against a session-cookie-based application?

    1. Encrypting cookies with a server key
    2. Setting the HttpOnly flag on cookies
    3. Validating all input on the server
    4. Requiring an anti-CSRF token Correct answer.

    Explanation

    An anti-CSRF token ties a state-changing request to the user's session and cannot be guessed by a cross-origin attacker. Cookie encryption and HttpOnly protect the cookie's contents from disclosure, not from being automatically attached to forged requests.

  164. Question 164Application Security

    Which technology runs inside the application runtime and can block attacks by inspecting calls in real time?

    1. Cloud web application firewall
    2. API gateway
    3. Static analysis scanner
    4. Runtime self-protection Correct answer.

    Explanation

    Runtime application self-protection instruments the application or its runtime and can observe and block malicious behavior during execution. A web application firewall sits at the network edge, static scanners operate before deployment, and gateways primarily route and police API traffic.

  165. Question 165Application Security

    In a microservices architecture, what is the primary security purpose of mutual TLS between services?

    1. Load balancing across instances
    2. Caching of responses at the edge
    3. Mutual authentication of both endpoints Correct answer.
    4. Compression of traffic between services

    Explanation

    Mutual TLS requires each side to present a valid certificate, so both the calling service and the receiving service authenticate each other. This supports zero-trust service-to-service communication and limits impersonation.

  166. Question 166Application Security

    An application accepts serialized objects from untrusted clients. Which design change most effectively removes the risk of insecure deserialization?

    1. Encrypting serialized objects in transit
    2. Using a data-only format with schema validation Correct answer.
    3. Adding integrity checks to serialized payloads
    4. Compressing serialized objects before sending

    Explanation

    Data-only formats with schema validation avoid instantiating arbitrary object types and therefore prevent gadget-chain exploitation. Controls such as integrity checks, encryption, and compression do not stop a malicious payload that is validly signed or encrypted.

  167. Question 167Application Security

    A service provider validates SAML assertions by checking only the signature on the Response element. Which attack does this permit?

    1. Replay of an expired assertion token
    2. XML external entity expansion attack
    3. Signature wrapping of the assertion Correct answer.
    4. Downgrade to unsigned SAML

    Explanation

    Signature wrapping exploits a mismatch between the element that was signed and the element the application later processes. The service provider must verify that the signature reference covers the exact assertion whose claims are consumed.

  168. Question 168Application Security

    A team adopting infrastructure as code wants to catch publicly accessible storage buckets before deployment. Which control fits best?

    1. WAF rules blocking bucket access
    2. DAST scanning of the running application
    3. RASP agents on production hosts
    4. IaC scanning in the CI pipeline Correct answer.

    Explanation

    Scanning infrastructure as code in the CI pipeline detects misconfigured resource definitions before they are provisioned. Runtime controls such as DAST, RASP, and WAFs act only after the infrastructure exists and cannot see the intended configuration.

  169. Question 169Incident Response & Resilience

    In the NIST incident response lifecycle, which phase immediately follows Detection and Analysis?

    1. Containment and recovery Correct answer.
    2. Post-incident activity
    3. Coordination and information sharing
    4. Preparation

    Explanation

    NIST SP 800-61 sequences incident response as Preparation, Detection and Analysis, Containment/Eradication/Recovery, and Post-Incident Activity. Containment and recovery therefore immediately follows detection and analysis. Post-incident review happens after recovery is complete.

  170. Question 170Incident Response & Resilience

    In an IaaS deployment, which party is normally responsible for monitoring the guest operating system for signs of compromise?

    1. The cloud provider's security operations team
    2. The data center colocation provider
    3. The hypervisor vendor
    4. The customer's security operations team Correct answer.

    Explanation

    In IaaS, the provider secures the physical infrastructure and hypervisor, while the customer secures the guest OS, applications, and identities. Monitoring guest OS compromise is therefore a customer responsibility.

  171. Question 171Incident Response & Resilience

    Which metric defines the maximum acceptable time a business process can be unavailable after a disruption?

    1. Service level objective (SLO)
    2. Recovery point objective (RPO)
    3. Recovery time objective (RTO) Correct answer.
    4. Mean time to detect (MTTD)

    Explanation

    RTO is the target elapsed time within which a business process must be restored after a disruption. RPO measures acceptable data loss, while MTTD measures detection speed.

  172. Question 172Incident Response & Resilience

    Which cloud action best preserves forensic evidence from a running virtual machine while keeping the instance available?

    1. Editing the instance's logs in place
    2. Terminating the instance and deleting its volumes
    3. Rebooting the instance repeatedly
    4. Snapshotting volumes and exporting logs Correct answer.

    Explanation

    Snapshots create point-in-time copies of attached volumes, and exporting logs preserves records before retention expires. Both actions support investigation without immediately destroying the running instance.

  173. Question 173Incident Response & Resilience

    Which activity belongs to the Preparation phase of incident response?

    1. Containing lateral movement across subnets
    2. Maintaining the IR plan and contact lists Correct answer.
    3. Writing the final post-incident report
    4. Eradicating malware from an infected host

    Explanation

    Preparation includes establishing plans, roles, contact lists, tooling, and training before an incident occurs. Eradication, containment, and post-incident reporting occur in later phases.

  174. Question 174Incident Response & Resilience

    A SaaS customer detects that an employee account is being used to exfiltrate data. Under the shared responsibility model, what should the customer do first?

    1. Activate its IR plan and disable the account Correct answer.
    2. Wait for the provider to notify all tenants
    3. Ask the provider to rewrite the application code
    4. Delete all tenant data to prevent further loss

    Explanation

    The SaaS provider manages application infrastructure, but the customer controls its user accounts, identities, and data access. The immediate containment step is to activate the customer IR plan and disable the compromised account.

  175. Question 175Incident Response & Resilience

    An analyst needs to stop malicious outbound traffic from a compromised cloud instance while preserving its state for investigation. Which action best meets both goals?

    1. Restrict network access and snapshot the disk Correct answer.
    2. Delete the instance's security group and logs
    3. Reboot the instance into safe mode
    4. Terminate the instance immediately

    Explanation

    A restrictive security group stops malicious traffic while a snapshot preserves a point-in-time disk image for analysis. Termination or reboot would destroy volatile evidence.

  176. Question 176Incident Response & Resilience

    Why is the log retention period a critical term to negotiate in a cloud incident response contract?

    1. It sets how far back logs allow reconstruction of activity Correct answer.
    2. It defines the provider's autoscaling limits
    3. It controls which encryption algorithm is used for stored data
    4. It sets the maximum number of users per tenant

    Explanation

    Log retention defines the historical window available for reconstructing attacker activity. Short provider retention can leave gaps unless logs are exported to customer-controlled storage.

  177. Question 177Incident Response & Resilience

    An application must survive the failure of a single data center. Which cloud architecture feature most directly provides this resilience?

    1. A larger instance type with more memory
    2. Multi-AZ deployment within one region Correct answer.
    3. Daily backups stored in the same rack
    4. A single-region deployment with one availability zone

    Explanation

    Availability zones are isolated data center groups within a region, so multi-AZ deployment tolerates one AZ failure. Regional disasters require multi-region design rather than only multi-AZ.

  178. Question 178Incident Response & Resilience

    Which characteristic most distinguishes cloud forensics from traditional on-premises forensics?

    1. Chain of custody is unnecessary in cloud cases
    2. Provider-held logs and no physical access Correct answer.
    3. Cloud networks never use segmentation
    4. Cloud data cannot be encrypted

    Explanation

    Cloud investigators usually cannot seize physical media and must depend on provider APIs, logs, and cooperation. This loss of physical control is a defining forensic challenge.

  179. Question 179Incident Response & Resilience

    A legal hold requires preservation of email stored in a SaaS productivity suite. What should the organization do?

    1. Instruct users to delete old messages to reduce storage
    2. Rely solely on the provider's standard backup rotation
    3. Suspend deletion policies and export the mailbox Correct answer.
    4. Change all user passwords and disable logging

    Explanation

    Legal hold requires affirmative preservation; normal deletion policies and backup rotation may expire data. Suspending deletion and exporting or placing the mailbox on hold preserves it.

  180. Question 180Incident Response & Resilience

    Why can destroying or revoking encryption keys be an effective containment action for compromised cloud object storage?

    1. It automatically restores the last clean backup
    2. It patches the vulnerability that allowed access
    3. It makes the encrypted data unreadable Correct answer.
    4. It prevents all network access to the storage service

    Explanation

    Destroying keys makes ciphertext unreadable, a technique called crypto-shredding, while leaving other tenants unaffected. It does not fix the vulnerability or restore data.

  181. Question 181Incident Response & Resilience

    A cloud provider discloses a hypervisor vulnerability affecting multi-tenant hosts. What is the customer's most appropriate response?

    1. Migrate all workloads to another provider immediately
    2. Assume no action is needed because the provider owns the host
    3. Patch the provider's hypervisor directly
    4. Patch guest systems and monitor provider remediation Correct answer.

    Explanation

    The provider remediates the hypervisor under infrastructure responsibility; the customer still patches guest OS and applications and watches for compromise. Immediate migration may be considered but is not the automatic first response.

  182. Question 182Incident Response & Resilience

    An investigation is hampered because a cloud provider retains platform logs for only seven days. Which change best improves future incident response resilience?

    1. Export logs to customer-controlled immutable storage Correct answer.
    2. Increase the instance size to generate more logs
    3. Move workloads to spot instances for lower cost
    4. Enable multi-factor authentication for administrators

    Explanation

    Exporting logs to customer-controlled immutable storage extends retention beyond provider defaults and preserves evidence. The other controls do not address the retention gap.

  183. Question 183Incident Response & Resilience

    A breach affects personal data stored in multiple countries with different breach notification laws. What is the best course of action?

    1. Delete the affected logs to limit legal exposure
    2. Wait until the cloud provider publicly discloses the incident
    3. Map each jurisdiction's notification rules with legal counsel Correct answer.
    4. Notify every affected person worldwide within 24 hours

    Explanation

    Breach notification triggers and timelines vary by jurisdiction, so legal and privacy counsel must map the obligations. A single global 24-hour notice may be wrong, and waiting for provider disclosure may miss deadlines.

  184. Question 184Incident Response & Resilience

    Which log source is most useful for determining which identity accessed a specific object in an object storage service?

    1. Physical data center badge reader records
    2. Virtual machine performance counters
    3. Access logs and audit trail data events Correct answer.
    4. Domain Name System query logs

    Explanation

    Object storage access logs and audit data events record object-level API calls and the requesting identity. They are the primary source for attribution of object access.

  185. Question 185Related Technologies & Strategies

    In a Functions-as-a-Service (FaaS) offering, how is patching of the underlying infrastructure handled?

    1. The customer patches the hypervisor but not the guest operating system.
    2. The customer patches the guest operating system and the language runtime.
    3. The provider patches servers and runtime; the customer owns function code. Correct answer.
    4. Patching is shared evenly between provider and customer at every layer.

    Explanation

    FaaS is a managed platform service: the provider operates the hosts, operating system, and language runtime, while the customer remains accountable for function code, third-party libraries, configuration, and the identities the function uses. This differs from IaaS, where guest patching sits with the customer.

  186. Question 186Related Technologies & Strategies

    Many IoT devices stay deployed in the field for a decade or more. Which security consequence does this most directly create?

    1. Devices with long lifecycles are unable to use TLS for transport encryption.
    2. Long service life with weak update mechanisms leaves known flaws unpatched. Correct answer.
    3. Long lifecycles force providers to rotate device credentials every 90 days.
    4. Devices already in the field cannot be monitored by network security tools.

    Explanation

    IoT risk is dominated by lifecycle: devices are often inexpensive, constrained, and lack signed firmware update channels, so vulnerabilities discovered after deployment persist for years. Longevity magnifies exposure because the device outlives the vendor's support and the organization's ability to replace it.

  187. Question 187Related Technologies & Strategies

    Which control best ensures that only reviewed container images can run in a production cluster?

    1. Storing Dockerfiles in a private source code repository.
    2. Scanning an image once at build time and never again.
    3. Deploying every image with the :latest tag for simplicity.
    4. Enforcing signature verification through an admission controller. Correct answer.

    Explanation

    Image signing plus admission control ties deployment to provenance: the cluster refuses any image whose signature cannot be verified against a trusted builder. Combined with continuous rescanning of the registry, this blocks tampered or unreviewed images, which a one-time scan cannot do.

  188. Question 188Related Technologies & Strategies

    A data lake ingests raw logs from many sources with widely differing sensitivity. Which practice best preserves least privilege?

    1. Copying all datasets into one shared analytics bucket for convenience.
    2. Relying on volume-level encryption as the sole security control.
    3. Classifying data at ingestion and enforcing fine-grained access at query time. Correct answer.
    4. Restricting access only at the reporting and dashboard layer.

    Explanation

    Least privilege in a data lake depends on knowing what each record is and restricting access at the point of use. Tagging at ingestion and enforcing attribute- or column-level policy at query time prevents analysts from seeing data they are not entitled to.

  189. Question 189Related Technologies & Strategies

    Why is encryption of long-lived sensitive data treated as urgent when planning for quantum computing?

    1. Ciphertext recorded today may be decrypted once a capable quantum computer exists. Correct answer.
    2. Grover's algorithm fully breaks AES-256 with a single query.
    3. Quantum key distribution is mandated for all public cloud storage.
    4. Quantum computers have already broken RSA-2048 in production attacks.

    Explanation

    The harvest-now-decrypt-later threat means an adversary can capture encrypted traffic or archives today and wait for a cryptographically relevant quantum computer. Data that must stay confidential for many years therefore needs post-quantum or hybrid algorithms before that machine exists.

  190. Question 190Related Technologies & Strategies

    An organization plans to store personal data on a permissioned blockchain. Which data-protection concern is most significant?

    1. Immutability guarantees that every stored record is accurate and trustworthy.
    2. Consensus algorithms encrypt every transaction end to end for all participants.
    3. Append-only records are hard to erase, conflicting with deletion obligations. Correct answer.
    4. Smart contracts are formally verified and cannot contain exploitable logic.

    Explanation

    Blockchain's append-only design is in direct tension with erasure and rectification rights: once a record is replicated across the ledger it cannot practically be removed. The usual mitigation is to keep personal data off-chain and store only pointers or hashes on-chain.

  191. Question 191Related Technologies & Strategies

    An attacker corrupts a portion of the training dataset so that a fraud detection model systematically misses certain transactions. What is this attack called?

    1. Model inversion
    2. Adversarial input evasion
    3. Data poisoning Correct answer.
    4. Membership inference

    Explanation

    Data poisoning targets the training phase, manipulating or injecting samples so the resulting model learns incorrect behavior. Attacks at prediction time against a finished model are instead called adversarial examples or evasion.

  192. Question 192Related Technologies & Strategies

    Which control most directly limits the impact of a lost or stolen corporate mobile device that accesses SaaS applications?

    1. Requiring longer passwords for the SaaS application.
    2. Device encryption plus remote wipe enforced by MDM/UEM. Correct answer.
    3. Blocking all mobile access to SaaS applications.
    4. SMS one-time codes as the only authentication factor.

    Explanation

    Encryption protects data at rest on the device, and remote wipe or selective wipe removes corporate data and tokens once the loss is reported. These two controls together address both the window before notification and the device's residual contents.

  193. Question 193Related Technologies & Strategies

    Where should pipeline credentials used to deploy to production be stored?

    1. Baked into the container image at build time.
    2. As plain environment variables in the repository's build file.
    3. In a shared team spreadsheet with restricted access.
    4. In a dedicated secrets manager injected at run time. Correct answer.

    Explanation

    Secrets managers issue short-lived, auditable credentials to the pipeline identity at run time, so the secret never enters source control or an image layer. Because images and repositories are widely readable and copied, any secret committed to them must be treated as compromised.

  194. Question 194Related Technologies & Strategies

    Which security challenge is most characteristic of edge computing deployments?

    1. Exposed sites and intermittent links complicate patching and monitoring. Correct answer.
    2. Edge hardware always sits inside the provider's data centre.
    3. Edge computing eliminates the need for a cloud shared responsibility model.
    4. Edge nodes remove the need to encrypt data in transit.

    Explanation

    Edge nodes often run in physically accessible, poorly controlled locations with unreliable backhaul, so centralized patching, logging, and monitoring assumptions break down. Designs must therefore include hardened devices, local buffering of telemetry, and secure remote attestation.

  195. Question 195Related Technologies & Strategies

    Which approach is most effective for reducing risky shadow IT use of SaaS?

    1. Requiring staff to sign an acceptable use policy at hire.
    2. Blocking all outbound HTTPS traffic from the corporate network.
    3. Relying on employees to self-report the tools they use.
    4. Discovering SaaS use with a CASB and offering approved tools. Correct answer.

    Explanation

    Shadow IT is driven by unmet needs, so discovery plus an approved alternative reduces it durably. Continuous CASB or API-based discovery surfaces unsanctioned services so risk can be assessed and sanctioned or blocked deliberately.

  196. Question 196Related Technologies & Strategies

    Employees paste customer records into a public generative AI chatbot. What is the primary security concern?

    1. The prompts are always encrypted with AES-256 before leaving the browser.
    2. The data may be retained or used for training outside organizational control. Correct answer.
    3. Prompts are always deleted immediately after the user's session ends.
    4. The model automatically refuses to process any sensitive content.

    Explanation

    Public consumer AI services typically retain inputs subject to their own terms, and data may be reviewed or used to improve models. Once pasted, the organization loses control of the copy, so acceptable use policy and enterprise tenants with no-retention terms are the mitigations.

  197. Question 197Related Technologies & Strategies

    In an event-driven serverless application, which control most directly mitigates injection attacks that reach backend data stores?

    1. Validating and schema-checking event payloads before any downstream use. Correct answer.
    2. Increasing the function's provisioned memory and timeout settings.
    3. Writing raw event bodies into the data store for later analysis.
    4. Relying solely on the provider's web application firewall.

    Explanation

    Events can originate from queues, buckets, and other services, not only from HTTP clients behind a WAF, so each function must treat event data as untrusted input. Strict schema validation and parameterized queries stop injected content before it is interpreted by a downstream system.

  198. Question 198Related Technologies & Strategies

    Which technique best supports aggregate analytics on a sensitive dataset while limiting re-identification of individuals?

    1. Base64-encoding the identifiers in each record.
    2. Replacing names with pseudonyms and keeping the mapping.
    3. Encrypting the dataset with a single shared key.
    4. Differential privacy with calibrated noise added to query results. Correct answer.

    Explanation

    Differential privacy bounds how much any single individual's presence can influence an output, so aggregate results remain useful while re-identification and inference are provably limited. Simple removal of names or reversible encodings leave enough quasi-identifiers for linkage attacks.

  199. Question 199Related Technologies & Strategies

    What risk does confidential computing primarily address in a public cloud?

    1. Network eavesdropping between availability zones.
    2. Weak authentication of human administrators.
    3. Loss of data at rest when a disk is decommissioned.
    4. Exposure of data in use to the provider's privileged software. Correct answer.

    Explanation

    Confidential computing uses hardware-based trusted execution environments so that memory is encrypted and isolated while a workload runs, keeping it opaque to the hypervisor and provider operators. It complements encryption at rest and in transit, which protect data in other states.

  200. Question 200Related Technologies & Strategies

    Which practice best embeds security into infrastructure-as-code pipelines?

    1. Reviewing templates manually once a year.
    2. Applying changes straight to production to save time.
    3. Storing secrets directly inside the templates.
    4. Scanning templates with policy-as-code checks before deployment. Correct answer.

    Explanation

    Policy-as-code evaluates every proposed change against rules for encryption, network exposure, logging, and approved services, so misconfigurations are blocked before resources exist. Automated evaluation also detects drift when templates are updated after initial review.

CCSK flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

NIST essential characteristics
Five criteria from NIST SP 800-145: on-demand self-service, broad network access, resource pooling, rapid elasticity, and measured service. A service must exhibit all five to qualify as cloud.
IaaS customer boundary
In IaaS the provider secures facilities, hardware, hypervisor, and physical network, while the customer is responsible for the guest operating system, middleware, runtime, applications, and data.
Cloud deployment models
Public, private, hybrid, and community. Hybrid joins two distinct models through data or application portability; community infrastructure is shared by organizations with common concerns and requirements.
CSA Cloud Controls Matrix
A cloud-specific control framework organized into domains and mapped to ISO 27001, PCI DSS, and NIST. It underpins the CAIQ self-assessment questionnaire and provider assurance comparisons.
SOC 2 Type II
An attestation report covering both the design and operating effectiveness of controls over a period, usually six to twelve months. Type I covers design only at a single point in time.
Compliance inheritance limits
A provider's certification covers only the provider's own scope of control. Customer-side obligations and the controls the customer operates are never inherited and must be evidenced separately.
Cloud Center of Excellence
A cross-functional team that defines cloud standards, guardrails, and reusable patterns centrally, while business units retain accountability for their own workloads under a federated governance model.
Data owner and custodian
The data owner is accountable for classification and access decisions; the custodian implements and operates the technical controls on the owner's behalf, often the cloud platform team.
OAuth 2.0
A delegated authorization framework that issues access tokens so a client can act on a resource owner's behalf. It is not an authentication protocol on its own.
SAML versus OpenID Connect
SAML 2.0 is an XML-based federation standard for browser single sign-on; OpenID Connect is a JSON and REST identity layer built on OAuth 2.0 that returns an ID token.
Short-lived credentials
Temporary tokens issued by a cloud security token service with a fixed expiry. They replace long-lived access keys and limit the damage from a leaked secret.
Least privilege
Granting only the permissions needed for a task and only for as long as needed, enforced in cloud with scoped roles, permission boundaries, and periodic entitlement review.
CloudTrail and Activity Logs
Control-plane audit logs record each API call in an account, capturing identity, source IP, timestamp, and outcome. They are the primary evidence source in cloud investigations.
VPC flow logs
Records metadata about IP traffic to and from network interfaces in a virtual network, including source, destination, port, protocol, bytes, and accept or reject action. Payloads are not captured.
Log immutability
Storing audit logs in write-once or object-lock storage prevents an attacker holding admin credentials from deleting evidence, preserving integrity for forensics and compliance obligations.
Security group versus NACL
Security groups are stateful instance-level virtual firewalls with allow-only rules. Network ACLs are stateless, subnet-level, and evaluate ordered rules that can both allow and deny.
Private endpoint
A provider-managed private connection that routes traffic to a service over the provider backbone instead of the public internet, reducing exposure and supporting residency requirements.
Infrastructure as code
Declarative templates that provision cloud resources repeatably under version control, enabling peer review, drift detection, and automated policy checks before anything is deployed.
Container escape
Because containers share the host kernel, a kernel or runtime vulnerability can let a process break isolation and reach other containers or the host itself.
Cloud Security Posture Management
A service that continuously assesses cloud configuration against benchmarks and policy, flagging exposures such as public storage buckets, open management ports, and overly permissive roles.
Immutable infrastructure
Servers and containers are never patched in place; a freshly built hardened image is deployed and the old one destroyed, which eliminates configuration drift and shrinks patch windows.
Envelope encryption
A data key encrypts the data, and a master key held in a key management service encrypts the data key. The wrapped data key is stored alongside the ciphertext.
Crypto-shredding
Destroying the encryption key rather than the data renders all ciphertext permanently unrecoverable, satisfying sanitization requirements for encrypted data at rest when no key copies remain.
BYOK versus HYOK
Bring Your Own Key means the customer generates or controls the key inside the provider's key service; Hold Your Own Key keeps the key material entirely outside the provider.
SAST versus DAST
Static application security testing inspects source code without executing it and finds flaws early; dynamic testing exercises a running application and finds runtime and configuration issues.
Secrets sprawl
Credentials hard-coded into source code, container images, or CI variables. It is mitigated with a secrets manager issuing short-lived tokens and automated commit and image scanning.
Metadata service SSRF
A server-side request forgery flaw lets an attacker query the instance metadata endpoint and steal the workload's temporary credentials, countered by hardened metadata access and egress filtering.
NIST incident response phases
Preparation; detection and analysis; containment, eradication, and recovery; and post-incident activity, as defined in NIST SP 800-61 and applied to cloud investigations.
RTO and RPO
Recovery Time Objective is the maximum acceptable downtime; Recovery Point Objective is the maximum acceptable data loss measured in time. Both derive from business impact analysis.
Cloud evidence preservation
Isolate a compromised instance by tightening its security group and snapshot its disks and memory before remediation, because terminating the instance destroys volatile forensic evidence.
Zero trust architecture
The NIST SP 800-207 model in which no entity is trusted because of network location; every request is authenticated and authorized per session by a policy decision point.
Function as a service
Event-driven serverless compute that scales automatically, is billed per invocation and execution duration, and is stateless with a maximum permitted execution time.

CCSK study guide

Cloud Computing Concepts & Architectures

Start with the NIST definitions

NIST SP 800-145 supplies the vocabulary the exam assumes. Five essential characteristics: on-demand self-service, broad network access, resource pooling, rapid elasticity, and measured service. Three service models (SaaS, PaaS, IaaS) and four deployment models (public, private, hybrid, community). You are rarely asked to recite these; you are asked to use them, because every question about who does what resolves through the service and deployment model in play.

The responsibility boundary moves

In IaaS the provider secures facilities, hardware, and the hypervisor; the customer owns the guest OS and everything above it. In PaaS the provider adds OS, runtime, and often middleware; the customer owns applications, data, and identity configuration. In SaaS the provider runs nearly everything, yet the customer still owns data classification, user lifecycle, sharing settings, and anything it configures.

Layers, not boxes

A useful mental model is a stack: physical, virtualization, network, storage, compute, application, data, and metadata. Each control is inherited, shared, or owned. Sketch that matrix before answering a responsibility question.

Traps

  • Private cloud does not by definition mean single-tenant; resource pooling and multi-tenancy remain normal.
  • Elasticity (automatic expansion and contraction with demand) is not the same as scalability.
  • Measured service describes metering capability, not a billing model.
  • Community cloud is shared by organizations with a common concern, not by the general public.

Rule of thumb

'The provider secures what it operates; you secure what you configure, put in, and let in.' If a control touches your data, your identities, or your configuration, assume you own it unless a contract says otherwise.

Cloud Governance, Risk, Audit, & Compliance

Governance in a rented world

Cloud governance is the set of policies, decision rights, and oversight mechanisms that keep cloud use aligned with business objectives. Practical artifacts include an enterprise cloud policy, acceptable-use rules, data classification, a service approval workflow, and defined exception handling. The frameworks to know by name: the CSA Cloud Controls Matrix (CCM) as a control catalog mapped to ISO 27001, PCI DSS, NIST, and others; the CAIQ as the self-assessment questionnaire that populates it; the STAR registry as the public repository of assessments; ISO/IEC 27017 for cloud-specific controls and 27018 for PII in public cloud; SOC 2 for service organizations; and FedRAMP for US federal authorizations.

Risk

Cloud risk assessment is ordinary risk assessment plus new dependencies: provider concentration risk, supply chain and subprocessor risk, data residency and sovereignty, lock-in and exit, and the risk that a control is asserted but not evidenced.

Audit

You will seldom audit a provider's datacenter. You inherit assurance through third-party reports, chiefly SOC 2 Type II (a period of time, tests operating effectiveness) versus Type I (a point in time, design only). Contracts should grant a right to audit, notification of material changes, and access to attestations.

Traps

  • SOC 2 is an attestation report, not a certification.
  • ISO 27001 certifies the provider's management system, not the security of your data.
  • The provider being compliant does not make you compliant; you still own your layer.
  • A CCM control marked shared does not mean the provider performs it.

Rule of thumb

Prefer evidence over assertion: for every inherited control, ask which document proves it and who signed it.

Organization Management

Cloud changes the org chart

Traditional separation, where a build team hands artifacts to an operations team, breaks when product teams self-provision. Expect questions on the Cloud Center of Excellence (CCoE), platform teams, cloud service owners, and federated security champions. The CCoE sets standards and enables delivery; it is not the security operations center. Distinguish accountability from responsibility: a provider may perform a control while your organization remains accountable for the outcome.

Roles to recognize

Cloud security architect, cloud security engineer, IAM administrator, FinOps analyst, procurement and vendor management, and data steward. Embedded security champions scale better than a central review gate, which becomes a bottleneck and is bypassed.

Culture and change management

The shift is from ticket-based provisioning to automated, policy-as-code provisioning with guardrails. That requires training, an acceptable-use policy, consequences for shadow IT, and change management for identity and third-party lifecycle. Personnel controls (screening, awareness training, termination procedures) appear in the CCM and are easy to overlook.

Traps

  • Assuming the provider's security team will notice or fix your misconfiguration.
  • Confusing the CCoE with the SOC, or governance with operations.
  • Choosing answers that position security as a blocker; the exam favors enabling guardrails.
  • Forgetting that organizational and human-resources controls are part of the control set.
  • Treating vendor management as procurement only, ignoring ongoing assurance.

Rule of thumb

If a decision cannot be expressed as an automated guardrail, it will eventually be bypassed under delivery pressure. Encode the policy, publish the exception process, and keep the human decision for exceptions only.

Identity & Access Management

Identity is the new perimeter

Cover the identity lifecycle (joiner, mover, leaver, with deprovisioning as the perennial exam answer), authentication, authorization, federation and single sign-on, multi-factor authentication, privileged access management, just-in-time elevation, secrets management, and role-based versus attribute-based access control. SAML and OpenID Connect handle authentication and federation; OAuth 2.0 is a delegated authorization framework. Cloud-native additions include entitlement analysis and permission boundaries.

Cloud-specific realities

Root, global administrator, and organization-owner accounts should be locked away with hardware MFA and break-glass procedures, never used for daily work. Workloads should use managed identities, instance profiles, or role assumption rather than embedded long-lived keys. A cloud policy statement is evaluated on principal, action, resource, and condition, with explicit deny typically winning. Separation of duties, scoped roles, and temporary credentials are the default posture.

Federation versus synchronization

Directory synchronization copies identities (and often password hashes) into the cloud. Federation trusts tokens and keeps authentication at the identity provider. When a question asks how to reduce credential sprawl or preserve a single authoritative directory, federation is the answer.

Traps

  • OAuth 2.0 is authorization; OpenID Connect is the authentication layer.
  • Encryption is not an access control; key access is the control.
  • Roles with reassuring names are often far broader than they sound.
  • MFA on the root account does not remove the need for least privilege.

Rule of thumb

Log in as root only to establish the break-glass process and the first administrator; everything else should be a federated role with conditions attached.

Security Monitoring

You cannot inherit logs

Monitoring in cloud means collecting, retaining, correlating, and alerting on both the management plane (API calls that create or change resources) and the data plane (access to data and network traffic). Representative native services: AWS CloudTrail, VPC Flow Logs, GuardDuty, and Config; Azure Activity Log, Monitor, Defender for Cloud, and NSG flow logs; Google Cloud Audit Logs, VPC Flow Logs, and Security Command Center. Concepts to recognize: SIEM ingestion and correlation, baselining normal behavior, anomaly detection, cloud security posture management, cloud workload protection, and data loss prevention on egress.

What to log

Identity events (console and API logins, failed authentication, privilege changes, key creation), network flow and DNS, storage access, and provider service health. Centralize into a separate, restricted, ideally immutable account or bucket with defined retention, and confirm retention outlives a plausible incident.

Traps

  • Audit logging is not always on, and data-plane or object-level logging usually must be deliberately enabled.
  • The provider monitors its own infrastructure, not the behavior inside your workloads.
  • Time matters: correlate on synchronized UTC timestamps.
  • Egress monitoring detects data leaving; it is not the same as blocking inbound traffic.

Rule of thumb

Before trusting any alert, answer three questions: is the log enabled in every account and region, is it tamper-resistant, and does a named person own the response?

Infrastructure & Networking

Design the virtual network first

Cloud networks are software-defined. The objects to know: VPC or VNet, subnets (public versus private), route tables, internet and NAT gateways, security groups, network ACLs, load balancers, DNS, and private connectivity options such as site-to-site VPN, AWS Direct Connect, Azure ExpressRoute, Google Cloud Interconnect, and private endpoints. Hybrid designs commonly use hub-and-spoke or transit-gateway topologies, and segmentation or microsegmentation for lateral movement control.

Security mechanics

Default-deny ingress, allow-listed egress where feasible, no default VPCs, and no management ports open to the internet. Keep management planes on private paths or behind bastion or zero-trust access. Web application firewalls filter HTTP-layer attacks; DDoS protection operates at the provider edge; TLS protects data in transit. Remember that provider APIs are internet-reachable, so protect them with identity and policy as well as network controls.

Traps

  • Security groups are stateful and support allow rules only; NACLs are stateless and support both allow and deny.
  • A private subnet needs a NAT gateway for outbound traffic, and that NAT is an egress path for your data.
  • VPC peering is non-transitive; transit gateways exist to solve scale.
  • Firewall, WAF, and DDoS protection cover different layers and are not interchangeable.

Rule of thumb

If a resource does not need to be reachable from the internet, it should have no public address and its route table should make that impossible rather than merely unlikely.

Cloud Workload Security

Protect the unit of compute

Workloads include virtual machines, containers, serverless functions, and managed platform runtimes. Unifying practices: hardened minimal base images, vulnerability and patch management, immutable infrastructure where change means redeployment rather than in-place patching, scanned and signed image registries, infrastructure-as-code scanning, runtime protection, and configuration hardening against a recognized benchmark.

Containers and serverless

Containers share the host kernel, so host hardening and orchestration control-plane security (role-based access in Kubernetes, admission control, network policy, secret handling) matter more than the container itself. Serverless shifts OS and runtime patching to the provider; you still own function permissions, dependency vulnerabilities, secrets, event injection, and timeouts. Ephemeral workloads must have logging and identity designed in from the start, because there is no machine to log into later.

Agents versus agentless

Agent-based tools provide runtime depth but require deployment and lifecycle management. Agentless approaches scan snapshots and API-visible configuration; easier to operate, but they see only what the API exposes. Mature programs use both.

Traps

  • Misconfiguration, not zero-days or hypervisor escape, is the dominant workload risk in cloud.
  • The provider does not patch your guest OS in IaaS or your container images anywhere.
  • Instance metadata services are a well-known attack path; require token-based access.
  • Immutable infrastructure means you do not log in and fix things by hand.

Rule of thumb

Treat every workload as disposable: build from code, scan the build, minimize what runs, and destroy rather than repair.

Data Security

Data is the asset

Think in lifecycle terms: create, store, use, share, archive, destroy. Control it with classification, inventory (including unknown or shadow data stores), ownership, and handling rules. Encryption applies in transit, at rest, and increasingly in use through confidential computing enclaves. The crux is key management: provider-managed keys, customer-managed keys (bring your own or hold your own), and external key stores or hardware modules. The more you hold the key, the more control and the more operational burden you accept.

Cloud-specific problems

Data dispersion and logical co-mingling with other tenants, residency and sovereignty obligations, default replication across regions, backup and snapshot copies, and orphaned data after deletion. Deleting a pointer, an object reference, or a virtual machine does not necessarily delete the underlying data. Understand retention windows and crypto-shredding, where destroying the key renders data unreadable.

Tooling

Data loss prevention, cloud access security brokers for sanctioned and unsanctioned SaaS visibility, tokenization and format-preserving encryption, database activity monitoring, and access analytics that show who actually touched the data.

Traps

  • Encryption without control of the key is weaker protection than it appears.
  • Deleting an object in a SaaS application may leave copies in version history, trash, or backups.
  • Data classification is the customer's duty in every service model.
  • The claim that a provider cannot access your data holds only if keys and access paths truly support it.

Rule of thumb

Ask three questions of any data store: who holds the key, who can be granted access, and where do the copies live?

Application Security

Shift left, guard the runtime

Secure development in cloud means threat modeling, secure design review, static and dynamic analysis, software composition analysis, secrets management, dependency scanning, and infrastructure-as-code scanning inside the pipeline. In cloud, the API is the application: inventory APIs, authenticate and authorize every call, validate input, rate-limit, and log. Classic risks persist, but cloud adds broken object-level authorization in APIs, application roles that are far broader than the code needs, and secrets committed to repositories or baked into environment variables.

Multi-tenancy in your own application

If you build SaaS, tenant isolation depends on your code: tenant identifiers enforced in every query, per-tenant keys, and shared caches or search indexes that do not leak across boundaries. This is the same pooled-resource problem the provider solves for you, one layer up.

Responsibility by model

In IaaS you harden OS and application. In PaaS you own code plus the configuration of the managed runtime: connection strings, feature flags, identity settings. In SaaS you own configuration, meaning sharing settings, SSO configuration, integrations, and the OAuth scopes you grant to third parties.

Traps

  • A web application firewall mitigates but does not fix vulnerable code.
  • Managed platform patching does not fix your dependency vulnerabilities.
  • A secret committed to a repository is compromised even if the repository is later made private.
  • Third-party OAuth consent can expose data with no exploitation at all.

Rule of thumb

Threat-model every trust boundary the application crosses: the user to the API, the API to cloud services, and your application to third-party SaaS.

Incident Response & Resilience

Plan for the cloud you have

The incident response lifecycle is preparation, detection and analysis, containment, eradication, recovery, and post-incident activity. Cloud changes the mechanics: you cannot unplug a disk, instances may be terminated by an attacker or by an autoscaler, and scope spans accounts, regions, and provider status events. Preparation includes pre-authorized forensic accounts and snapshot permissions, named provider contacts, and log retention that outlives the incident.

Forensics challenges

Shared tenancy means no access to the physical layer, and volatile evidence disappears the moment an instance stops. Investigators rely on API and audit logs, disk snapshots, in-guest memory capture, and provider cooperation under contract. Jurisdiction and chain of custody still apply, and data may sit in several regions at once.

Resilience

Recovery time and recovery point objectives drive design. Distinguish backups from snapshots, from multi-zone replication, from multi-region architectures. Multi-zone failover delivers availability, not backup: deletion, corruption, and ransomware replicate happily. Provider disaster recovery covers provider infrastructure, not your data or configuration. Test restores and rehearse under realistic conditions.

Traps

  • High availability is not a backup strategy.
  • Snapshots must be consistent and stored where you can actually restore them.
  • The provider's incident response plan is not your incident response plan.
  • Auto-scaling and self-healing can undo containment actions.

Rule of thumb

Contain by identity and network first (revoke credentials, apply deny policies, isolate the subnet), then preserve evidence, then rebuild from trusted code.

Related Technologies & Strategies

The surrounding landscape

The exam expects literate awareness rather than depth of adjacent topics: big data and analytics pipelines, where data gravity and residency bite; IoT, with hard device-identity problems, weak update paths, and edge processing; containers and orchestration; serverless; edge computing; machine learning, including training-data exposure and model theft; blockchain, whose immutability conflicts with erasure obligations; confidential computing; and post-quantum planning.

Strategies that keep appearing

Cloud access security brokers discover and govern shadow SaaS, operating in forward or reverse proxy mode, API mode, or log-based mode. SASE and SD-WAN converge networking and security at the edge. Zero trust means no implicit trust from network location: verify explicitly, apply least privilege, assume breach. Microsegmentation, security as code, shadow-IT discovery, and treating the CI/CD pipeline as a production system all recur.

How to answer

These questions usually test whether you can map a new technology onto the shared responsibility model. Ask: which abstraction layer is this, who operates it, what identity does it carry, where does its data live, and what happens when it fails or is decommissioned?

Traps

  • Proxy-based CASB requires traffic to route through the vendor; API-mode CASB sees only what the SaaS API exposes.
  • IoT devices rarely support MFA or endpoint agents, so compensate with network segmentation.
  • Blockchain immutability conflicts with data-deletion rights under privacy regulation.
  • Zero trust is an architecture and policy model, not a purchasable product.

Rule of thumb

New technology, old question: whose layer is it, and where is the data?

CCSK cheat sheet

Quick reference

Cloud Computing Concepts & Architectures

  • NIST SP 800-145 defines five essential characteristics: on-demand self-service, broad network access, resource pooling, rapid elasticity, and measured service; a service must exhibit all five to be 'cloud'.

  • Three service models: SaaS (provider runs the whole application stack), PaaS (provider runs OS, runtime and middleware; customer deploys code and data), IaaS (provider runs virtualization and hardware; customer runs the guest OS upward).

  • Four deployment models: public, private, hybrid, and community; hybrid is two distinct models joined by data or application portability, not simply 'private plus public'.

  • Shared responsibility shifts with the model: IaaS customer owns OS and above; PaaS customer owns application, configuration, and data; SaaS customer owns data, identities, and endpoint use.

  • Resource pooling implies multitenancy, so tenant isolation depends on hypervisor separation, network segmentation, and correct tenant-ID validation in every request.

  • Elasticity creates lifecycle risk: rapid provisioning produces orphaned volumes, snapshots, and identities that persist after workloads are destroyed unless de-provisioning is automated.

  • NIST names five cloud actors: consumer, provider, auditor, broker, and carrier; the broker aggregates and mediates services, which adds a supply-chain trust hop.

  • ISO/IEC 17788 provides the standardized vocabulary and ISO/IEC 17789 the reference architecture; CSA's reference model aligns governance, operations, and service-model layers.

Quick reference

Cloud Governance, Risk, Audit, & Compliance

  • Risk = likelihood x impact; cloud-specific risks include shared tenancy, provider lock-in, data residency, supply-chain dependency, provider insolvency, and loss of exit capability.

  • CSA Cloud Controls Matrix (CCM) is a cloud-specific control set that maps to ISO 27001, NIST, and PCI DSS; the CAIQ is its matching self-assessment questionnaire.

  • The CSA STAR registry publishes provider assurance at Level 1 (self-assessment), Level 2 (third-party audit), and Level 3 (continuous audit-based certification).

  • SOC 2 Type I attests control design at a single point in time; Type II attests design and operating effectiveness across a period, typically six to twelve months.

  • Compliance is never fully inherited: a provider's certificate covers only the provider's scope, while customer-side controls (identity governance, data classification, configuration) must be evidenced separately.

  • Audit rights must be contracted explicitly: right to audit, access to reports, penetration-test summaries, notification of material changes, and breach notification windows.

  • Policy hierarchy: policy (mandatory intent) > standard (specific requirements) > procedure (steps) > guideline (advisory); only the first three are enforceable.

  • Jurisdictional data location is a legal fact, not a technical control; replication, backups, CDN caching, and support access can move or expose data across borders.

  • KRIs measure risk exposure (e.g., count of public storage buckets) and KPIs measure control performance (e.g., percent of assets scanned); both should feed the governing body.

Quick reference

Organization Management

  • A Cloud Center of Excellence centralizes standards, guardrails, and reusable patterns while business units retain accountability for their own workloads under a federated model.

  • The shared responsibility split must be documented per service model and per control, with exactly one Accountable owner in the RACI for each control.

  • Typical cloud accountability split: security owns policy; platform engineering owns guardrails and landing zones; application teams own configuration of their own workloads.

  • Role definitions to assign explicitly: cloud service owner, data owner, data custodian, security architect, and incident commander.

  • Skill shift: hardware and network operations decline while infrastructure as code, policy as code, API automation, and dependency management increase.

  • Tagging and showback/chargeback link cost accountability to security accountability; untagged resources are simultaneously a budget and an inventory gap.

  • Third-party management becomes continuous monitoring of assurance artifacts rather than one-time onboarding, because provider configurations and subprocessors change.

  • Change management must extend to IaC pipelines; peer review and automated policy gates supplement or replace parts of traditional change advisory approval.

  • Human-factor controls (phishing resistance, secrets handling, privileged access hygiene) remain the highest-yield awareness investments in cloud environments.

Quick reference

Identity & Access Management

  • Cloud IAM policy evaluation order: explicit deny overrides explicit allow, and anything not explicitly allowed is implicitly denied.

  • Federation uses SAML 2.0 or OpenID Connect to assert identity across trust domains; OAuth 2.0 is delegated authorization and is not an authentication protocol by itself.

  • SCIM standardizes automated provisioning and, critically, deprovisioning between an identity provider and SaaS applications.

  • Prefer short-lived credentials from a security token service over long-lived access keys; static keys are the most common source of cloud credential leaks.

  • Least privilege is enforced with scoped roles or policies, permission boundaries, and periodic entitlement reviews; unused-permission analysis tools identify over-grant.

  • Privileged access requires MFA (hardware-backed for root/global admin), just-in-time elevation, and session logging; root accounts should not be used for daily operations.

  • Zero trust replaces network-location trust with per-request authentication and authorization based on identity, device posture, and context.

  • De-provisioning is incomplete unless it also removes API keys, service accounts, federated trust relationships, and OAuth consent grants.

  • Separation of duties applies to cloud too: the identity that deploys infrastructure should not also be able to alter audit logging or key policies.

Quick reference

Security Monitoring

  • Core log planes: control plane (AWS CloudTrail, Azure Activity Log, GCP Cloud Audit Logs), network (VPC flow logs), workload agents, and application logs.

  • Management-plane logs are the highest-value cloud telemetry because API abuse by a compromised principal looks like ordinary administrative traffic.

  • Aggregate logs to a dedicated security account or subscription with restricted write access; store in object-lock/WORM storage to preserve forensic integrity.

  • SIEM pipeline: normalization, enrichment, correlation rules, alert triage, and tuning; MITRE ATT&CK for cloud supplies technique-level detection mapping.

  • Cloud-native detection services (GuardDuty, Security Hub, Defender for Cloud, Security Command Center) complement rather than replace customer correlation.

  • Default retention (often 90 days hot) usually fails both regulatory minimums and investigation needs; set retention from legal and incident-response requirements.

  • Baseline normal API behavior per identity, then alert on new regions, new user agents, disabled logging, key creation, and mass data egress.

  • Providers cannot see inside customer workloads; guest-level and application telemetry must be collected and shipped out by the customer.

  • Clock synchronization matters: correlate events across services and providers using NTP-synchronized UTC timestamps.

Quick reference

Infrastructure & Networking

  • Security groups are stateful, instance-level virtual firewalls with allow-only rules; network ACLs are stateless and subnet-level with ordered allow and deny rules.

  • Standard VPC pattern: public subnets only for ingress tiers, private subnets for workloads and data, outbound traffic through NAT or an inspecting proxy with logging.

  • Private connectivity options include VPC peering, transit gateways, private endpoints/PrivateLink, and dedicated interconnect; use them instead of routing sensitive traffic over the internet.

  • Plan non-overlapping CIDR ranges before deployment; overlapping address space blocks peering, merger, and hybrid connectivity later.

  • Software-defined networking makes all network controls API-driven, so misconfiguration rather than hardware failure is the dominant network risk.

  • Layered protections: provider DDoS absorption at layers 3/4, security groups and NACLs for segmentation, WAF for layer 7 HTTP(S), and identity-aware proxies for application access.

  • Zero trust networking uses micro-segmentation and per-request authorization; no implicit trust is granted from being inside a VPC or subnet.

  • Infrastructure as code requires version control, peer review, drift detection, and policy-as-code guardrails (OPA, AWS Config rules, Azure Policy).

  • Harden with CIS Benchmarks for the specific OS and service versions in use; immutable images reduce drift compared with in-place configuration.

Quick reference

Cloud Workload Security

  • Workload categories differ in control surface: VMs (full OS control), containers (shared kernel), serverless functions (code, role, and event source only), and managed PaaS runtimes.

  • Containers share the host kernel, so isolation depends on namespaces, cgroups, seccomp/AppArmor profiles, and non-root execution; a kernel exploit escapes every container on the host.

  • Scan images for vulnerabilities at build time and continuously in the registry, sign them, and enforce admission control so only trusted images run.

  • Immutable infrastructure means rebuild rather than patch: golden images must be regenerated on each patch cycle and the old ones retired.

  • For serverless, the provider secures the runtime; the customer secures function code, its IAM role, third-party dependencies, and event-source permissions.

  • CSPM finds configuration drift and misconfiguration; CWPP protects running workloads; both should feed the same risk register rather than separate queues.

  • Ephemeral workloads lose local logs on termination, so telemetry must be shipped off-instance in near real time.

  • Secrets must never be baked into images, user data, or environment files; use a secrets manager or workload identity with short-lived tokens.

  • Agentless scanning extends vulnerability coverage to instances that cannot host agents, but agent-based telemetry remains necessary for runtime detection.

Quick reference

Data Security

  • Data lifecycle: create/classify, store, use, share, archive, destroy; each phase has distinct controls and different accountable owners.

  • Encryption states: at rest (AES-256 or equivalent), in transit (TLS 1.2+), and in use (confidential computing, enclaves, or privacy-preserving computation).

  • Envelope encryption: a data key encrypts the data and a KMS master key encrypts the data key; the wrapped data key is stored alongside the ciphertext.

  • Key custody models progress from provider-managed to customer-managed (BYOK) to hold-your-own-key (HYOK) to client-side encryption, each shifting control and operational burden to the customer.

  • Crypto-shredding - destroying the key so ciphertext is unrecoverable - is an accepted sanitization method for encrypted data, but only if no plaintext copies or key backups remain.

  • Classification drives control selection: labels enable DLP, CASB policy enforcement, and access decisions based on data sensitivity.

  • Data residency (where data is stored) and data sovereignty (which laws apply) are distinct; replication, backups, and support access can violate residency assumptions.

  • Sanitization of shared storage should follow NIST SP 800-88; verify the provider's documented media destruction process for the storage class used.

  • Tokenization and masking shrink compliance scope (notably PCI DSS) by removing sensitive values from systems that do not need them.

Quick reference

Application Security

  • Shift-left toolchain: threat modeling at design, SAST on source, SCA on dependencies, DAST on running environments, and IAST/RASP for runtime protection.

  • Secrets sprawl is the top cloud application risk; enforce a secrets manager with short-lived tokens and automate commit and image scanning for credentials.

  • The CI/CD pipeline is a high-value target: harden build agents, isolate runners per trust level, sign artifacts, and give deploy identities least privilege.

  • IaC scanning catches insecure cloud configuration (public buckets, open security groups, unencrypted volumes) before resources are created.

  • API security requires authentication, authorization per object, rate limiting, schema validation, and an inventory that catches shadow and deprecated endpoints.

  • Serverless attack surface is event-driven: validate event payloads, restrict function role permissions, and treat event sources as untrusted input.

  • Broken access control on object storage and SSRF to the instance metadata service are cloud-specific additions to the OWASP Top 10 baseline.

  • SaaS risk concentrates in OAuth consent grants, third-party integrations, and external sharing settings that bypass the application's own access checks.

  • Secure SDLC gates must be automated and blocking in the pipeline; documented manual review processes degrade without enforcement.

Quick reference

Incident Response & Resilience

  • NIST SP 800-61 phases: preparation; detection and analysis; containment, eradication, and recovery; post-incident activity.

  • Cloud IR lacks physical access and media seizure; use provider APIs for disk snapshots, memory capture where supported, and log export as the evidence path.

  • Preserve before remediating: isolate an instance by tightening its security group and snapshot volumes, because termination destroys volatile evidence permanently.

  • Pre-negotiate provider escalation: contract SLAs, support tier contacts, law-enforcement request procedures, and the provider's own incident notification commitments.

  • RTO is maximum tolerable downtime and RPO is maximum tolerable data loss; both derive from a business impact analysis, not from technical convenience.

  • Multi-AZ addresses availability within a region; multi-region addresses disaster recovery; the two are not substitutes for each other.

  • Backups must be tested for restore and protected with object lock or air-gapped copies to resist ransomware deleting the primary and replicas.

  • Failover tests must include identity provider, DNS, and key management - the most common single points of failure in cloud DR designs.

  • Breach notification clocks (e.g., GDPR 72 hours) start at awareness, not at conclusion of the investigation, so parallel legal escalation is required.

Quick reference

Related Technologies & Strategies

  • Containers virtualize the operating system and share the host kernel; virtual machines virtualize hardware and each carry a full guest OS.

  • Function as a service is event-driven, auto-scaling, stateless, billed per invocation and duration, and bounded by a maximum execution time.

  • Big data platforms shift the control focus to data-lake access control, lineage, metadata cataloging, and encryption of both raw and derived datasets.

  • IoT devices are constrained, long-lived, and poorly patchable; security depends on unique device identity, gateway segmentation, and lifecycle replacement planning.

  • DevOps is culture plus automation plus measurement plus sharing; DevSecOps inserts automated security gates into that pipeline rather than adding a separate review stage.

  • Zero trust architecture per NIST SP 800-207 separates the policy decision point from the policy enforcement point and evaluates every request on context.

  • SASE and SD-WAN converge networking and security at the edge, moving inspection from a central data center to the user and workload location.

  • Cryptographic agility is a design requirement: algorithm and key length choices must be replaceable without re-architecting data protection.

CCSK key terms

NIST Cloud Computing Definition

A widely used model describing cloud computing as on-demand network access to a shared pool of configurable computing resources that can be provisioned and released rapidly with minimal provider interaction or management effort.

Related: Essential Characteristics, Cloud Service Models
Essential Characteristics

The five traits NIST uses to define cloud: on-demand self-service, broad network access, resource pooling, rapid elasticity, and measured service. A service must show all five to count as cloud computing.

Related: NIST Cloud Computing Definition, Multitenancy
Multitenancy

A single instance of software or infrastructure serves multiple customers, with each tenant's data and activity logically isolated. Pooling resources this way lowers provider cost but makes isolation controls and misconfiguration the main security concern.

Related: Essential Characteristics, Shared Responsibility Model
Cloud Service Models

The three delivery categories - Infrastructure as a Service, Platform as a Service, and Software as a Service - defined by how much of the stack the provider manages versus the customer. Higher-level models shift more security responsibility to the provider.

Related: Shared Responsibility Model, Cloud Deployment Models
Cloud Deployment Models

The four arrangements for hosting cloud services - public, private, hybrid, and community - describing who owns the infrastructure and who shares it. The choice affects control, cost, compliance obligations, and exposure to other tenants.

Related: Multitenancy, Cloud Service Models
Shared Responsibility Model

The division of security duties between a cloud provider and its customer. The provider secures the underlying facilities and platform; the customer secures data, identities, configuration, and whatever the service model leaves in their control.

Related: Cloud Service Models, Cloud Controls Matrix
Cloud Governance

The policies, decision rights, and oversight mechanisms an organization uses to direct cloud adoption - approving services, setting standards, managing cost and risk, and holding teams accountable across business units.

Related: Cloud Controls Matrix, Cloud Center of Excellence
Cloud Controls Matrix (CCM)

The Cloud Security Alliance's catalog of security control objectives mapped to industry standards, used to assess cloud providers and build control frameworks. It gives a common vocabulary for comparing provider assurances.

Related: STAR Registry, Cloud Governance
STAR Registry

CSA's public repository of provider self-assessments via the Consensus Assessments Initiative Questionnaire, third-party audits, and certifications. Buyers use it for preliminary due diligence before requesting more detailed evidence.

Related: Cloud Controls Matrix (CCM), FedRAMP
FedRAMP

A U.S. government program that authorizes cloud services for federal use through standardized security assessments, continuous monitoring, and a shared repository of authorization packages. Many commercial buyers treat it as evidence of maturity.

Related: STAR Registry, Cloud Governance
Data Residency

Rules and practices that keep data physically stored or processed within specified geographic or legal boundaries. It stems from law, contract, or policy and must be checked against where a provider replicates and backs up.

Related: Data Classification, Cloud Deployment Models
Cloud Center of Excellence

A cross-functional team that sets cloud standards, builds shared landing zones and guardrails, and coaches product teams. It centralizes expertise early while pushing day-to-day delivery into business units.

Related: Cloud Governance, DevSecOps
Exit Strategy

A documented plan for moving data, workloads, and identities off a provider - covering export formats, deletion confirmation, transition timelines, and dependencies on proprietary services - agreed before the contract is signed.

Related: Cloud Governance, Data Remanence
Federated Identity

An arrangement where a trusted identity provider authenticates users and passes assertions to cloud applications, so accounts and credentials are managed centrally rather than per service. It relies on standards such as SAML and OpenID Connect.

Related: SAML, Privileged Access Management
SAML

An XML-based standard for exchanging authentication and authorization assertions between an identity provider and a service provider. It is common for browser single sign-on to enterprise cloud applications.

Related: Federated Identity, Zero Trust
Privileged Access Management

Tools and processes that control, vault, and record high-privilege accounts such as cloud administrators and root users. It limits standing access, enforces approval workflows, and creates an audit trail of administrative actions.

Related: Federated Identity, Secrets Management
Secrets Management

Secure storage and controlled retrieval of credentials such as API keys, tokens, and database passwords, usually through a dedicated vault with rotation and short-lived issuance. It removes hard-coded secrets from code and configuration.

Related: Privileged Access Management, IaC Scanning
Cloud Access Security Broker (CASB)

A policy enforcement point placed between users and cloud services that provides visibility into sanctioned and unsanctioned app use, plus controls for authentication, encryption, data loss prevention, and malware.

Related: Data Loss Prevention (DLP), Cloud Security Posture Management (CSPM)
SIEM

A platform that collects, normalizes, correlates, and alerts on log and event data from many sources. In cloud environments it must ingest provider control-plane logs, workload logs, and identity events to detect attacks.

Related: Cloud Security Posture Management (CSPM), Incident Response Plan
Cloud Security Posture Management (CSPM)

A tool that continuously compares cloud configuration against policy and benchmarks, flagging risky settings such as public storage buckets or overly permissive roles. It works from provider APIs rather than host agents.

Related: Cloud Workload Protection Platform (CWPP), Cloud Controls Matrix (CCM)
Virtual Private Cloud (VPC)

A logically isolated network within a provider's shared infrastructure where you define subnets, routing, and address ranges. It is the boundary for many network controls and peering relationships.

Related: Security Group, Microsegmentation
Security Group

A stateful, resource-level firewall rule set that permits or denies traffic to cloud resources. Because rules are allow-lists attached to resources, overly broad rules are a frequent finding in assessments.

Related: Virtual Private Cloud (VPC), Microsegmentation
Microsegmentation

Granular network control that isolates individual workloads or groups from one another rather than relying only on a perimeter. It limits lateral movement after a compromise and is often enforced through policy tied to workload identity.

Related: Security Group, Zero Trust
Private Interconnect

A dedicated private network link between an organization's data center and a cloud provider, bypassing the public internet. It improves predictability and reduces exposure but does not by itself encrypt traffic.

Related: Virtual Private Cloud (VPC), Encryption at Rest and in Transit
Cloud Workload Protection Platform (CWPP)

A security product focused on protecting running workloads - virtual machines, containers, and serverless functions - through vulnerability scanning, behavioral monitoring, and runtime protection. It complements posture tools that assess configuration.

Related: Cloud Security Posture Management (CSPM), Container Image Scanning
Container Image Scanning

Automated inspection of container images for known vulnerabilities, embedded secrets, and policy violations before deployment. Scanning in the build pipeline and registry catches problems before they reach production.

Related: Cloud Workload Protection Platform (CWPP), DevSecOps
Encryption at Rest and in Transit

Protecting stored data with encryption keys and protecting data moving across networks with protocols such as TLS. Both are baseline controls, but key custody and configuration determine how much protection they actually provide.

Related: Key Management Service (KMS), Hardware Security Module (HSM)
Key Management Service (KMS)

A managed service for creating, storing, rotating, and controlling access to cryptographic keys. It centralizes key policy so that encryption across many cloud services can be audited and revoked consistently.

Related: Hardware Security Module (HSM), Encryption at Rest and in Transit
Hardware Security Module (HSM)

A tamper-resistant hardware device that generates and stores keys and performs cryptographic operations inside a protected boundary. Cloud providers offer them as managed or dedicated services for high-assurance key custody.

Related: Key Management Service (KMS), Encryption at Rest and in Transit
Data Classification

Assigning data to sensitivity categories so that handling, access, retention, and protection requirements follow the label. Effective classification is a prerequisite for scoping encryption, monitoring, and loss-prevention controls.

Related: Data Loss Prevention (DLP), Data Residency
Data Loss Prevention (DLP)

Controls that detect and block sensitive data leaving approved boundaries, using pattern matching, labels, or context. In cloud, DLP is applied at endpoints, email, SaaS APIs, and storage services.

Related: Data Classification, Cloud Access Security Broker (CASB)
Data Remanence

Residual copies of data left on storage media after deletion or reassignment. In cloud, customers must rely on provider sanitization commitments, crypto-shredding by destroying keys, and contractual deletion terms.

Related: Key Management Service (KMS), Exit Strategy
Shift Left

Moving security activities earlier in the development lifecycle - into design, coding, and build - so defects are cheaper to fix and controls are built in rather than audited afterward.

Related: SAST and DAST, IaC Scanning
SAST and DAST

Two application testing approaches: static analysis inspects source or binaries without running the program, while dynamic analysis attacks a running application. They find different flaw classes and are usually combined.

Related: Shift Left, DevSecOps
IaC Scanning

Examining infrastructure-as-code templates for insecure defaults, missing controls, and policy violations before deployment. It prevents misconfigurations - a leading cloud risk - from being provisioned in the first place.

Related: Shift Left, Cloud Security Posture Management (CSPM)
Incident Response Plan

A documented set of roles, escalation paths, communication channels, and procedures for detecting, containing, eradicating, and recovering from a security incident. It must account for provider dependencies and shared responsibilities.

Related: Cloud Forensics, RTO and RPO
Cloud Forensics

Collecting and preserving evidence from cloud systems, where logs and snapshots may be volatile, provider-controlled, or spread across regions. It requires pre-arranged provider cooperation and legally defensible acquisition procedures.

Related: Incident Response Plan, Data Residency
RTO and RPO

Recovery Time Objective is the maximum acceptable downtime; Recovery Point Objective is the maximum acceptable data loss measured in time. Both drive backup frequency, replication choices, and failover architecture.

Related: Incident Response Plan, Private Interconnect
Zero Trust

A security model that grants access per request based on identity, device, and context rather than assuming a trusted network. It replaces perimeter trust with continuous verification and least privilege.

Related: Microsegmentation, Federated Identity
DevSecOps

Integrating security practices, automation, and ownership into DevOps delivery so controls run continuously in pipelines and teams share accountability for security outcomes rather than handing off to a separate team.

Related: Shift Left, IaC Scanning, Cloud Center of Excellence

CCSK study plan

8 weeks · About 6 hours per week

Week 1

Cloud Computing Concepts & Architectures

  • Diagram the three service models and mark precisely where the customer's security responsibility ends in each.

  • Compare the major deployment models and list which governance and operational controls migrate to the provider in each case.

  • Distinguish shared responsibility from shared fate and apply the distinction to compute, storage, network, identity, and data.

  • Assess how multitenancy, resource pooling, and elasticity change isolation boundaries and what compensating controls each demands.

  • Outline the CSA security guidance domain structure and map the twelve exam domains onto it in your own words.

Your checkpoint

A one-page responsibility matrix with rows for compute, storage, network, identity, and data, and columns for IaaS, PaaS, and SaaS, rebuilt from memory and scored at 90% or better against source material.

Week 2

Cloud Governance & Risk, Audit, & Compliance

  • Draft a cloud governance charter that assigns decision rights for procurement, data classification, and provider onboarding.

  • Build a risk register containing ten cloud-specific risks, each scored for likelihood and impact with a stated treatment decision.

  • Map control framework domains to a sample audit scope and separate inherited controls from controls the customer must evidence.

  • Differentiate assurance types by stating exactly what each proves and what it does not: independent attestation report, certification audit, and self-attestation questionnaire.

  • Explain how to negotiate audit rights, regulator access, and evidence substitution in a cloud service agreement.

Your checkpoint

A ten-row cloud risk register, each row tagged to a control framework domain and a treatment decision, plus a one-paragraph rationale for the top three risks.

Week 3

Organization Management & Identity & Access Management

  • Define a cloud security organizational structure that separates policy ownership, platform ownership, and workload ownership.

  • Write role charters for three roles and identify the overlap and gap between them.

  • Distinguish authentication, authorization, and federation, and state the enforcement point for each.

  • Design a least-privilege role plus an attribute-based policy for a multi-tenant workload, describing conditions and boundaries.

  • Compare federation protocols and contrast the risk profile of long-lived API keys against short-lived brokered tokens.

Your checkpoint

An identity and access design sheet for one workload covering identities, roles, trust relationships, credential lifetime, and the MFA enforcement point.

Week 4

Identity & Access Management & Security Monitoring

  • Enumerate the log sources required for cloud detection across control plane, data plane, identity, network, and SaaS audit trails, and set a retention decision for each.

  • Design a detection pipeline from collection through normalization, correlation, alerting, and ticketing.

  • Specify five high-value alerts with thresholds, data sources, and the response each must trigger.

  • Separate what the provider supplies from what the customer must build for monitoring in each service model.

Your checkpoint

A monitoring coverage matrix mapping five log sources to five alerts, with an owner and retention period per row, plus a written statement of the three largest remaining blind spots.

Week 5

Infrastructure & Networking & Cloud Workload Security

  • Design segmented cloud networking with subnets, stateful versus stateless filtering, and controlled egress.

  • Select between hub-and-spoke and flat topologies for a stated constraint set and justify the choice.

  • Harden a workload image through baseline definition, patch cadence, immutability, and secrets injection at runtime.

  • Explain container and serverless isolation boundaries and identify where customer controls begin in each.

  • Choose between agent-based and agentless workload protection for two contrasting requirement sets.

Your checkpoint

A combined network and workload hardening checklist of fifteen items, each tagged to a domain 7 or 8 control and paired with a verification method.

Week 6

Data Security & Application Security

  • Classify data into four tiers and specify encryption requirements at rest, in transit, and in use for each tier.

  • Design a key hierarchy with explicit key custody, rotation, and the tradeoffs of customer-managed versus provider-managed keys.

  • Contrast tokenization, format-preserving encryption, and field-level encryption by use case and failure mode.

  • Apply secure development practices to a cloud pipeline: secrets scanning, infrastructure-as-code scanning, dependency scanning, and release gating.

  • Explain tenant data isolation options in a shared application and the failure modes of each.

Your checkpoint

A data protection specification for one data store covering classification, encryption points, key custody, retention and deletion, and the full access path.

Week 7

Incident Response & Resilience & Related Technologies & Strategies

  • Build an incident response plan adapted to cloud realities: provider notification paths, evidence acquisition order, and containment through identity and network controls.

  • Assign customer versus provider responsibilities at each incident response phase.

  • Define RTO and RPO targets and select a matching resilience pattern from backup-restore through active-active multi-region.

  • Evaluate at least three related technologies for their security impact on the domains already studied.

  • Execute a restore of sample data and record the actual measured time to restore.

Your checkpoint

A one-page incident response runbook for a compromised-credential scenario plus a restore test log with measured RTO compared against target.

Week 8

Cloud Computing Concepts & Architectures & Risk, Audit, & Compliance

  • Re-rank all twelve domains by practice accuracy and restudy only the three weakest.

  • Rebuild the week 1 responsibility matrix from memory and reconcile every difference against the original.

  • Complete two full-length timed assessments and log every miss with a root cause category.

  • Produce a decision aid that maps common business requirements to specific cloud security controls.

Your checkpoint

A weakness log that closes every practice-assessment miss with a corrected explanation, plus a final full-length assessment score at or above 80%.

Milestones

Week 4: Score at least 80% on a self-authored 25-question quiz covering domains 1 through 4, and reconcile the week 1 responsibility matrix against source material with no unexplained gaps.

Week 6: Complete the data protection specification and score at least 80% on a 20-question quiz covering domains 5 through 10, with any weak domain added to the week 8 restudy list.

Week 8: Achieve at least 80% on a final full-length timed assessment with a fully closed weakness log and a rebuilt responsibility matrix that matches the week 1 version.

CCSK exam day

Your checklist

Government-issued photo ID

Bring/display an unexpired government-issued photo ID whose name matches your exam registration exactly. Confirm accepted ID types with the provider before exam day.

Exam appointment confirmation

Have the registration/confirmation email or portal login ready so you can launch the exam and reach support if needed.

Reliable internet connection

Use a wired or stable high-speed connection; run a speed test and know how to tether or switch networks if the primary connection drops.

Working webcam and microphone

Test both before exam day; have a backup headset/mic if possible. Confirm exact device requirements with the provider.

Charged laptop with backup power

Fully charge the device and keep the charger plugged in; close battery-draining background apps.

Quiet, private, well-lit room

Clear your desk of papers, phones, and notes unless the provider explicitly permits them; ensure the room is private for the full session.

Clean desktop and browser

Close all other applications and tabs, disable notifications, and stop any screen-sharing or recording software before launching the proctored session.

Water in a clear container

If permitted, keep a clear glass or bottle nearby. Confirm the provider's policy on drinks and food.

Scratch paper and pen

Some proctored exams prohibit scratch paper; do not assume it is allowed. Confirm with the provider and remove it if not permitted.

Reference material plan (if open book)

If the exam is open book, have approved reference documents open in a permitted format and know how to search them quickly. Confirm the open-book policy and allowed materials with the provider.

Support contact details

Save the provider's exam support phone number or chat link in case of technical failure during check-in or the test.

Early check-in

Log in 15–30 minutes before your appointment to complete identity verification and system checks.

Plan the day

Final 72 hours

Re-read the exam blueprint and your weak-domain notes; take one timed practice set to check pacing, then stop heavy new study and focus on recall.

Final 72 hours

Run a full technical dry run: webcam, microphone, browser, internet speed, and login flow. Fix anything that fails now, not on exam day.

Final 24 hours

Confirm appointment time and time zone, review ID and room rules, prepare your workspace, and get a full night's sleep. Avoid cramming new material.

Morning of the exam

Eat, hydrate, and do a short review of key frameworks and definitions only. Re-test your camera, mic, and connection.

30 minutes before start

Clear your desk, silence your phone, close all apps, and begin check-in early so proctoring setup does not eat into exam time.

During the test

Read each question fully, flag uncertain items, and keep a steady pace across the 60-question, 90-minute format; return to flagged questions only after completing the first pass.

During the test

If a technical issue occurs, stop and contact proctoring support immediately rather than continuing blindly; note the time and issue.

Rules to check

  • Follow all proctoring instructions; unauthorized materials, phones, or people in the room can void your result. Confirm the exact permitted-items list with the provider.

  • Keep your face and workspace visible; do not leave the camera view or the room without following the proctor's instructions.

  • Use only the ID and materials the provider allows; name mismatches are a common cause of check-in failure.

  • Assume no scheduled breaks unless the provider states otherwise; plan hydration and restroom visits before the session. Confirm break policy in advance.

  • The exam is timed; unanswered or flagged questions still count against your time, so manage pacing deliberately.

  • Scoring and passing standards are set by the provider; confirm the current pass mark and retake policy rather than relying on secondhand figures.

  • Do not attempt to copy, share, or reconstruct exam content; doing so violates the provider's terms and can lead to credential revocation.

  • If you need accommodations or a reschedule, arrange it with the provider before exam day; last-minute requests may not be honored.

CCSK FAQ

How many CCSK practice questions are available?

The practice set has 200 CCSK questions. Each one includes the choices, the correct answer, and an explanation, and the set is free to read with no signup.

How many CCSK flashcards are available?

The flashcard set has 32 CCSK cards. Each card shows the term and its definition.

Which topics does the CCSK study material cover?

The CCSK study material has 11 sections, including Cloud Computing Concepts & Architectures, Cloud Governance, Risk, Audit, & Compliance, and Organization Management.

What format facts does the catalog list for the CCSK exam?

The catalog lists a duration of 120 minutes and these item types: multiple choice, open-book, and online for the CCSK exam.