Fabric DP-700 Practice Test

Implementing Data Engineering Solutions Using Microsoft Fabric (DP-700)

Fabric DP-700 overview

  • 200 practice questions
  • 33 flashcards
  • 3 study guide topics
  • 38 key terms

What you'll study

  • Implement and manage an analytics solution33%
  • Ingest and transform data33%
  • Monitor and optimize an analytics solution34%

Microsoft Certified: Fabric Data Engineer Associate is an intermediate-level Microsoft certification earned by passing exam DP-700: Implementing Data Engineering Solutions Using Microsoft Fabric. It targets data engineers who already have subject matter expertise with data loading patterns, data architectures, and orchestration processes, and who are skilled at manipulating and transforming data by using SQL, PySpark, and Kusto Query Language (KQL). The job role involves ingesting and transforming data, securing and managing an analytics solution, and monitoring and optimizing an analytics solution, working alongside analytics engineers, architects, analysts, and administrators.

The exam assesses three skills areas, each published as a range: Implement and manage an analytics solution (30–35%), Ingest and transform data (30–35%), and Monitor and optimize an analytics solution (30–35%). Within those areas the study guide spans workspace configuration, lifecycle management and deployment pipelines, security and governance, orchestration with pipelines and notebooks, batch and streaming ingestion, transformation patterns, error resolution, and performance tuning. You have 100 minutes; the exam is proctored and may include interactive components.

Preparation should combine Microsoft Learn self-paced training with real Fabric workloads: build lakehouses, pipelines, eventstreams, and warehouse objects, then practice diagnosing failures and tuning queries. Use the exam sandbox to learn the interface and the current practice assessment to gauge readiness. This page is independent study material and is not affiliated with or endorsed by Microsoft.

Common mistakes

Building dev, test, and prod artifacts in a single workspace and granting every team member the Admin role 'so nothing blocks them'.

Candidates optimize for getting the lab-style tasks done quickly and never practise the workspace-role model, so they forget that Admin/Member can manage access and that a single workspace makes promotion and rollback impossible to demonstrate.

What to do instead

Create one workspace per environment, promote content with a deployment pipeline, and assign the least-privileged role that works (Admin/Member for owners, Contributor for developers, Viewer for consumers). Reserve item-level permissions and sensitivity labels for data that needs stricter control.

Promoting items through a deployment pipeline without creating deployment rules, so the promoted semantic model or dataflow still points at the dev Lakehouse, SQL endpoint, or connection.

The deployment succeeds and the item looks healthy, so candidates assume the work is finished; they do not realize connections and parameters are not rewritten automatically unless a rule is defined for that stage.

What to do instead

Define deployment rules per stage for data source rules on semantic models and for parameters on pipelines, dataflows, and notebooks. Then validate by running the promoted item against the target environment, not just by checking that deployment reported success.

Defaulting every workload to a Lakehouse (or every workload to a Warehouse) instead of choosing the store from the workload's shape.

Candidates learn one item type well during preparation and reuse it everywhere, missing that T-SQL-heavy relational modelling, streaming/time-series, and file-based Spark processing are served by different Fabric stores.

What to do instead

Decide from requirements: Warehouse for relational T-SQL, multi-table transactions, and BI over structured data; Lakehouse for Spark, Delta engineering, and files/semi-structured data; Eventhouse/KQL database for high-volume streaming and time-series querying; mirrored/shortcut sources when data must stay in place.

Assuming a Direct Lake semantic model always reads Delta tables directly and never falls back, and never checking the model's mode after a refresh.

Direct Lake looks like 'Import but faster', so candidates skip the guardrails; when a table exceeds row/cardinality limits or uses an unsupported schema construct, the model silently falls back to DirectQuery and performance changes without any visible error.

What to do instead

Design Delta tables to stay within Direct Lake guardrails (reasonable column cardinality, supported types, well-defined Delta schema), check the mode reported after refresh, and capture Direct Lake fallback events through monitoring/logging so a fallback is detected rather than discovered as a performance complaint.

Copying data with a pipeline into the Lakehouse when the source is already ADLS Gen2/OneLake, instead of creating a shortcut.

Candidates equate 'ingest' with 'copy', so they pay for duplicate storage and repeated movement, and they miss questions about zero-copy access and where the data actually lives.

What to do instead

Use shortcuts when the requirement is read access to data that should stay in its source system with no duplication. Use Copy activity or notebooks when you must land a transformed, typed, or governed copy (schema mapping, type conversion, deduplication).

Reloading full source tables on every pipeline run instead of implementing an incremental pattern.

Full loads always produce correct results in a small test dataset, so the mistake is invisible until volume and cost matter; candidates then answer the design question with a full-refresh copy or a notebook that overwrites the Delta table.

What to do instead

Carry a watermark (last-modified or a high-water-mark column) and load only new/changed rows, then apply MERGE INTO on the Delta target to upsert. Use Dataflow Gen2 incremental refresh or a Copy activity with a watermark query when no notebook is involved.

Using Dataflow Gen2 with the default (non-staged) settings for large-volume transformations that would be better handled by a notebook or stored procedure.

Dataflows Gen2 are the easiest low-code path, so candidates use them for everything and overlook that transformation runs in the Mashup engine and that staging to a Lakehouse changes how and where the compute happens.

What to do instead

Reserve Dataflows Gen2 for low-code, moderate-volume transformations and connector-heavy ingestion. For large or complex transformations, use notebooks on Spark or T-SQL in the Warehouse; when a dataflow must handle bigger volumes, enable staging so the heavy work is pushed down to the Lakehouse.

Writing streaming or frequently-appended data into Delta tables and never compacting them, leaving thousands of tiny Parquet files.

The write succeeds and queries work on a small demo, so the small-file problem stays hidden; candidates do not connect slow downstream reads and inflated metadata to the append pattern.

What to do instead

Schedule OPTIMIZE (with V-Order as appropriate) on frequently written tables, run VACUUM to remove unreferenced files while respecting the retention period, and for high-rate streaming consider Eventhouse instead of landing raw events in Delta first.

Diagnosing slow or throttled workloads by intuition instead of reading the capacity metrics, then 'fixing' performance by re-running jobs at a different time.

Fabric compute is shared and smoothed over time, so a single slow run does not reveal whether the cause is a background job being throttled, an interactive query being delayed, or genuine query inefficiency.

What to do instead

Use the Capacity Metrics app to identify the item and operation consuming the most capacity units, check for throttling and which stage it affects, and correlate with the Monitoring hub. Then act on the specific consumer (optimize the query/model, move heavy background work, or right-size the capacity) rather than rescheduling blindly.

Tuning a slow notebook by rewriting the transformation code without first opening the Spark UI or the run's details to see where the time went.

Spark failure symptoms look like code problems, but skew, shuffle, small files, and executor sizing appear only in stage and task metrics; candidates who skip that step rewrite the wrong part of the job.

What to do instead

Open the Spark UI from the notebook run or Monitoring hub, identify the slow stage, and check task skew, shuffle read/write, spill, and input file sizes. Then apply the matching fix - repartition or broadcast for skew and joins, compaction for small files, caching for reused DataFrames - and for Warehouse T-SQL use Query Insights instead.

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

Fabric DP-700 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 1Implement and manage an analytics solution

    A workspace administrator must be able to add and remove workspace members and delete the workspace. Which workspace role provides these capabilities?

    1. Admin Correct answer.
    2. Contributor
    3. Viewer
    4. Member

    Explanation

    Workspace roles are Admin, Member, Contributor, and Viewer. Only Admin can manage workspace membership and delete the workspace, in addition to all item permissions.

  2. Question 2Implement and manage an analytics solution

    Which configuration enables a workspace to host Fabric items such as lakehouses and notebooks?

    1. Assigning Power BI Premium Per User licenses to all users
    2. Assigning the workspace to a Fabric capacity Correct answer.
    3. Connecting the workspace to an Azure SQL database
    4. Assigning Power BI Pro licenses to all users

    Explanation

    Fabric items are created in workspaces that are backed by a capacity, such as a Fabric F SKU, a Power BI Premium P SKU, or a Fabric trial capacity. Per-user Pro or PPU licenses alone do not enable Fabric item creation.

  3. Question 3Implement and manage an analytics solution

    When data is loaded into a managed table in a Fabric lakehouse, where and in what format is the table stored?

    1. CSV files in the Files folder
    2. Parquet files in the Files folder without a transaction log
    3. JSON documents in the Tables folder
    4. Delta Lake files in the Tables folder Correct answer.

    Explanation

    Lakehouse managed tables are stored as Delta Lake tables under the Tables folder of the lakehouse. The Files folder is intended for files that are not managed as tables.

  4. Question 4Implement and manage an analytics solution

    What is the primary characteristic of a OneLake shortcut?

    1. It references data in another location without copying it Correct answer.
    2. It converts external data into CSV files
    3. It copies source data into the lakehouse on a recurring schedule
    4. It creates a cached replica that must be refreshed

    Explanation

    A shortcut is a OneLake item that points to data stored elsewhere, such as another lakehouse, ADLS Gen2, or Amazon S3. Reads go to the target data, so no copy is created.

  5. Question 5Implement and manage an analytics solution

    In a medallion architecture, what is the main purpose of the gold layer?

    1. Serve curated, aggregated data for BI Correct answer.
    2. Archive deleted records for compliance
    3. Store raw data exactly as received
    4. Hold deduplicated but unvalidated data

    Explanation

    The gold layer contains business-level, aggregated, and curated data that is ready for reporting and analytics. Raw data belongs to bronze, and cleaned but not yet modeled data belongs to silver.

  6. Question 6Implement and manage an analytics solution

    Which Fabric feature is designed to promote items from a development workspace through test and production workspaces?

    1. OneLake shortcuts
    2. Git-based source control integration
    3. Data pipelines
    4. Deployment pipelines Correct answer.

    Explanation

    Deployment pipelines define stages, such as development, test, and production, and deploy item definitions between the workspaces assigned to those stages. Git integration provides version control but does not manage staged promotion.

  7. Question 7Implement and manage an analytics solution

    Which source control providers can a Fabric workspace connect to for Git integration?

    1. Only on-premises Team Foundation Version Control
    2. Azure DevOps Repos and GitHub Correct answer.
    3. Only GitHub
    4. Only Azure DevOps Repos

    Explanation

    Fabric Git integration supports connecting a workspace to either an Azure DevOps repository or a GitHub repository. On-premises TFVC is not a supported provider.

  8. Question 8Implement and manage an analytics solution

    Which endorsement level indicates that an item has been reviewed and approved by an authorized reviewer?

    1. Promoted
    2. Verified
    3. Certified Correct answer.
    4. Endorsed

    Explanation

    Certified is the higher endorsement level and requires a reviewer with permission to certify items. Promoted indicates that the owner considers the item ready to use but does not require review.

  9. Question 9Implement and manage an analytics solution

    Which T-SQL construct enforces row-level security in a Fabric warehouse?

    1. A column-level GRANT statement
    2. A security policy with a predicate function Correct answer.
    3. A stored procedure that checks SUSER_SNAME()
    4. A view that filters rows with a WHERE clause

    Explanation

    Row-level security in a Fabric warehouse is implemented with CREATE SECURITY POLICY, which applies an inline table-valued function as a predicate to a table. Views and stored procedures can be bypassed and do not enforce RLS by themselves.

  10. Question 10Implement and manage an analytics solution

    A varchar column uses the default dynamic data masking function, and a user has no UNMASK permission. What does the user see in that column?

    1. 0
    2. XXXX Correct answer.
    3. NULL
    4. 1900-01-01

    Explanation

    Default masking replaces string values with XXXX, numeric values with 0, and date/time values with 1900-01-01. Users who need the real values must be granted UNMASK.

  11. Question 11Implement and manage an analytics solution

    A scheduled pipeline must read from Azure Data Lake Storage Gen2 when no user is signed in. Which identity should be granted access to the storage account?

    1. The tenant's Power BI service account
    2. The workspace identity Correct answer.
    3. The capacity administrator's user account
    4. The pipeline author's user account

    Explanation

    A workspace identity is a Microsoft Entra service principal associated with the workspace, and Fabric items can use it for automated access to external storage. Granting the workspace identity permission on the storage account allows scheduled runs without user credentials.

  12. Question 12Implement and manage an analytics solution

    A pipeline must copy 20 source tables with different names into a lakehouse. What is the most maintainable design?

    1. Use a single copy activity with a wildcard query
    2. Create 20 pipelines triggered by one schedule
    3. Create 20 separate copy activities with hard-coded table names
    4. Use a ForEach activity with a parameterized copy activity Correct answer.

    Explanation

    A ForEach activity can iterate over a list of table names and pass each name as a parameter to a single copy activity. This avoids duplicating pipeline logic and makes changes easier.

  13. Question 13Implement and manage an analytics solution

    Which Fabric item is best suited for analysts who need low-code transformations with Power Query connectors before writing results to a lakehouse?

    1. Spark notebook
    2. Dataflow Gen2 Correct answer.
    3. Mirrored database
    4. KQL queryset

    Explanation

    Dataflow Gen2 provides a Power Query low-code experience with many connectors and supports a lakehouse or warehouse as an output destination. Notebooks require code, KQL querysets query KQL databases, and mirrored databases replicate source data.

  14. Question 14Implement and manage an analytics solution

    A Spark notebook is attached to a lakehouse. Which action makes a DataFrame available as a managed table through the lakehouse SQL analytics endpoint?

    1. Writing the DataFrame to the Files folder as CSV
    2. Saving the DataFrame as a Delta table in the Tables folder Correct answer.
    3. Calling df.cache()
    4. Registering a temporary view with createOrReplaceTempView

    Explanation

    Managed tables appear in the lakehouse SQL analytics endpoint when Delta data is written under the Tables folder, for example with saveAsTable. Temporary views are scoped to the notebook session and are not exposed through the endpoint.

  15. Question 15Implement and manage an analytics solution

    A solution must ingest thousands of events per second from Azure Event Hubs and route them to a KQL database with minimal code. Which Fabric item should be used?

    1. Notebook with a scheduled trigger
    2. Data pipeline with a copy activity
    3. Dataflow Gen2
    4. Eventstream Correct answer.

    Explanation

    Eventstream is the Fabric real-time ingestion item that can read from Event Hubs and route events to destinations such as a KQL database, lakehouse, or derived stream. Pipelines and dataflows are batch-oriented, and scheduled notebooks introduce latency.

  16. Question 16Implement and manage an analytics solution

    What does Fabric mirroring provide for a source such as Azure SQL Database?

    1. A one-time copy that must be refreshed manually
    2. Query federation that leaves the data in the original source system
    3. A writable copy that replaces the source
    4. Continuous replication of changes into OneLake as read-only Delta tables Correct answer.

    Explanation

    Mirroring continuously replicates source changes into OneLake, where the data is stored as Delta tables that are read-only. The source system remains the system of record and no ETL is required.

  17. Question 17Implement and manage an analytics solution

    Before renaming a warehouse table, a developer wants to see which reports and downstream items depend on it. Which feature should they use?

    1. Deployment pipelines
    2. Capacity metrics app
    3. Lineage view Correct answer.
    4. Sensitivity labels

    Explanation

    The lineage view shows the upstream and downstream dependencies of an item in the workspace. It helps identify the reports and other items affected by a change.

  18. Question 18Implement and manage an analytics solution

    Users report slow interactive queries and rejected operations during peak hours. Which tool helps confirm that the workspace is exceeding its capacity compute limits?

    1. OneLake file explorer
    2. Deployment pipeline history
    3. Git integration history
    4. The Capacity Metrics app Correct answer.

    Explanation

    The Microsoft Fabric Capacity Metrics app shows capacity unit consumption, throttling, and the effect of smoothing over time. It is the primary tool for investigating capacity pressure.

  19. Question 19Implement and manage an analytics solution

    A sensitivity label is applied to a lakehouse. What happens to downstream Fabric items that reference the lakehouse?

    1. They are automatically blocked from being shared
    2. They are moved to a labeled workspace
    3. They inherit the label when it is more restrictive than their own Correct answer.
    4. They keep the workspace default label and ignore the lakehouse label

    Explanation

    Sensitivity labels propagate from upstream items to downstream items, and the most restrictive label is retained. This keeps protection consistent across the data estate.

  20. Question 20Implement and manage an analytics solution

    A user must be able to view one report in a workspace without access to the other items. What should be done?

    1. Assign the user the Viewer workspace role for the workspace
    2. Assign the user the Contributor workspace role for the workspace
    3. Share the report item directly with the user Correct answer.
    4. Add the user to the workspace Admin role

    Explanation

    Sharing an individual item grants access to that item without granting workspace-wide access. Workspace roles apply to all items in the workspace.

  21. Question 21Implement and manage an analytics solution

    A lakehouse contains a folder named HR with sensitive files, and several teams have Contributor access to the workspace. How can access to the HR folder be limited to a specific security group?

    1. Apply a sensitivity label to the HR folder that restricts access
    2. Create a OneLake data access role scoped to the HR folder Correct answer.
    3. Move the folder to the Files section
    4. Create a workspace role assignment for the group

    Explanation

    OneLake data access roles grant or restrict access at the folder level inside a lakehouse, independent of workspace roles. Assigning the security group to a role scoped to the HR folder restricts other workspace users from reading that folder.

  22. Question 22Implement and manage an analytics solution

    In a deployment pipeline, the source connection string differs between the test and production stages. What should be configured?

    1. A parameter rule that overrides the value at deployment
    2. A deployment schedule per stage
    3. A data source rule that changes the connection Correct answer.
    4. A separate pipeline for each stage

    Explanation

    Deployment pipeline rules let you change values during deployment. A data source rule substitutes the connection used by an item in the target stage, while parameter rules override parameter values.

  23. Question 23Implement and manage an analytics solution

    A Delta table is frequently filtered on a customer ID column, and queries scan many small files. Which action improves read performance by colocating related rows?

    1. Enabling the table for mirroring
    2. Converting the table to CSV
    3. Running OPTIMIZE with ZORDER BY on the column Correct answer.
    4. Running VACUUM with a one-hour retention period

    Explanation

    OPTIMIZE compacts small files, and ZORDER BY colocates rows with similar values in the same files. This reduces the amount of data read when queries filter on the Z-ordered column.

  24. Question 24Implement and manage an analytics solution

    Which condition causes a Direct Lake semantic model to fall back to DirectQuery at query time?

    1. The model is refreshed on a schedule
    2. The model contains a measure written in DAX
    3. The lakehouse uses Delta Parquet tables
    4. The model uses row-level security Correct answer.

    Explanation

    Direct Lake reads Delta files directly, but some features cannot be served that way, including row-level security defined in the semantic model, which causes a fallback to DirectQuery. DAX measures and Delta tables are normal and do not cause fallback.

  25. Question 25Implement and manage an analytics solution

    A partner organization in another Microsoft Entra tenant needs read access to a specific lakehouse folder, and the data must not be copied. Which Fabric capability should be used?

    1. A data pipeline that copies the folder
    2. External data sharing Correct answer.
    3. A shortcut to the partner's storage
    4. A mirrored database

    Explanation

    External data sharing lets a tenant share OneLake data with users in another tenant without copying it. The recipient accesses the shared data through a shared item in their own tenant.

  26. Question 26Implement and manage an analytics solution

    An F64 Fabric capacity is used only during business hours, and the administrators want to stop compute billing overnight. What should they do?

    1. Enable autoscale on the capacity.
    2. Change the capacity SKU to F2.
    3. Pause the capacity in the Azure portal. Correct answer.
    4. Remove the workspace assignment from the capacity each evening.

    Explanation

    Pausing a Fabric capacity stops compute billing until it is resumed, while data remains in OneLake. Resuming restores the same capacity size and workspace assignments. Storage charges continue while the capacity is paused.

  27. Question 27Implement and manage an analytics solution

    A user must be able to create and edit items in a workspace but must not be able to grant other users access to that workspace. Which workspace role should be assigned?

    1. Contributor Correct answer.
    2. Admin
    3. Member
    4. Viewer

    Explanation

    The Contributor role can create, edit, and delete workspace items but cannot manage workspace access. Member and Admin roles include the ability to add users to the workspace. Viewer is read-only.

  28. Question 28Implement and manage an analytics solution

    An organization wants to group workspaces by business unit so that governance can be delegated to owners in each unit. Which Fabric feature should be used?

    1. Deployment pipelines
    2. Endorsement
    3. Sensitivity labels
    4. Domains Correct answer.

    Explanation

    Domains group workspaces and items by business area and allow domain admins to govern their own area. A workspace can be assigned to a single domain. This supports federated governance across a large tenant.

  29. Question 29Implement and manage an analytics solution

    A lakehouse must expose Delta tables that already reside in an Azure Data Lake Storage Gen2 account, and the data must not be duplicated. What should be created?

    1. An OneLake shortcut that points to the storage location Correct answer.
    2. A dataflow Gen2 that loads the tables into the lakehouse
    3. A mirrored database for the storage account
    4. A data pipeline that copies the tables into the lakehouse

    Explanation

    A shortcut is a reference to data held in another location, so Fabric reads the files in place without copying them. Copy activities, dataflows, and mirroring all materialize a second copy of the data.

  30. Question 30Implement and manage an analytics solution

    A team needs a relational T-SQL surface that supports full DML and multi-statement transactions over tables stored in OneLake. Which Fabric item should they use?

    1. A warehouse Correct answer.
    2. A KQL database
    3. A lakehouse SQL analytics endpoint
    4. A mirrored database

    Explanation

    A Fabric warehouse is a full T-SQL engine that supports INSERT, UPDATE, DELETE, and multi-statement transactions over Delta tables in OneLake. Its SQL analytics endpoint counterpart exposes read-only queries over lakehouse tables.

  31. Question 31Implement and manage an analytics solution

    Which write-time Parquet optimization stores Delta files in a layout that Fabric compute engines can read faster because it applies sorting and compression in a specific format?

    1. Z-Order
    2. Optimize write
    3. VACUUM
    4. V-Order Correct answer.

    Explanation

    V-Order is applied when Parquet files are written and produces a layout optimized for Fabric engines such as Direct Lake and Spark. Z-Order is a data-skipping layout applied during OPTIMIZE, and VACUUM removes unreferenced files.

  32. Question 32Implement and manage an analytics solution

    Which Delta operation permanently deletes data files that are no longer referenced by the table log and are older than the retention threshold?

    1. FSCK REPAIR TABLE
    2. A log checkpoint
    3. VACUUM Correct answer.
    4. OPTIMIZE

    Explanation

    VACUUM removes files that the transaction log no longer references once they exceed the retention threshold, which defaults to seven days. Running it with a shorter threshold reduces the window in which time travel is possible.

  33. Question 33Implement and manage an analytics solution

    A developer needs to read a Delta table exactly as it existed three versions ago. Which capability should be used?

    1. Query the table with time travel by specifying the version. Correct answer.
    2. Create a shortcut to the table's folder.
    3. Restore the workspace from a backup.
    4. Query the SQL analytics endpoint with NOLOCK.

    Explanation

    Delta keeps a transaction log of table versions, so a query can specify a version or timestamp to read historical data. Time travel is bounded by log retention and by VACUUM removing old files.

  34. Question 34Implement and manage an analytics solution

    Schema support is enabled for a new lakehouse. Which statement describes how its tables are organized by default?

    1. Tables are created in a schema named dbo unless another schema is specified. Correct answer.
    2. Tables cannot be created until an administrator creates a schema manually.
    3. All tables are placed in a single schema that cannot be renamed or extended.
    4. Tables are created in a schema named default until one is renamed.

    Explanation

    When schemas are enabled, the lakehouse includes a dbo schema that is used when no schema is named. Additional schemas can then be created to group tables by subject area or layer.

  35. Question 35Implement and manage an analytics solution

    A report developer needs read-only T-SQL access to the Delta tables in a lakehouse. Which endpoint provides this capability?

    1. The workspace Git remote endpoint
    2. The eventhouse query endpoint
    3. The lakehouse OneLake DFS endpoint
    4. The lakehouse SQL analytics endpoint Correct answer.

    Explanation

    Every lakehouse exposes a SQL analytics endpoint that serves read-only T-SQL over its Delta tables. The OneLake DFS endpoint is a storage API used by engines and tools rather than an interactive T-SQL surface.

  36. Question 36Implement and manage an analytics solution

    A query in a Fabric warehouse must join a table that resides in a second warehouse in the same workspace. How should the second table be referenced?

    1. Through a linked server defined on the warehouse.
    2. With a four-part name that includes a server address.
    3. With a three-part name in the form database.schema.table. Correct answer.
    4. By copying the table into the first warehouse before querying.

    Explanation

    Fabric warehouses support cross-database queries within the same workspace by using three-part names. No linked server or server address is involved because the databases share the workspace's SQL endpoint.

  37. Question 37Implement and manage an analytics solution

    Analysts must be able to create new reports from an existing semantic model but must not be able to change the model itself. Which permission should be granted on the semantic model?

    1. Write
    2. Build Correct answer.
    3. Reshare
    4. Read

    Explanation

    Build permission lets a user create new reports and other content on top of a semantic model without editing it. Read only allows viewing existing reports that use the model.

  38. Question 38Implement and manage an analytics solution

    In a Fabric warehouse, a user must be able to query some columns of the dbo.Customers table but must not see the Email column. Which approach enforces this requirement?

    1. Grant SELECT on the permitted columns only. Correct answer.
    2. Add the user to the workspace Viewer role.
    3. Deny SELECT on the table to the user.
    4. Apply a dynamic data masking rule to the Email column.

    Explanation

    Granting SELECT on individual columns implements column-level security, so the user can query only those columns. Masking keeps the column visible and returns masked values instead of hiding the column entirely.

  39. Question 39Implement and manage an analytics solution

    What is created automatically when a new lakehouse is added to a workspace?

    1. A KQL database
    2. A default semantic model for the lakehouse Correct answer.
    3. A mirrored database
    4. A deployment pipeline

    Explanation

    A lakehouse is created together with a SQL analytics endpoint and a default semantic model so its Delta tables can be consumed by Power BI. The default model is generated from the lakehouse tables.

  40. Question 40Implement and manage an analytics solution

    A Spark notebook requires a specific third-party Python library and custom Spark pool settings. Which Fabric item should supply these to the notebook?

    1. A KQL database
    2. An environment Correct answer.
    3. A data pipeline
    4. A warehouse

    Explanation

    An environment item stores Spark runtime settings, libraries, and pool configuration that notebooks and Spark job definitions can use. Attaching the environment applies those settings whenever the Spark session starts.

  41. Question 41Implement and manage an analytics solution

    A Spark job must run against a lakehouse on a schedule without an interactive notebook session. Which Fabric item is designed for this?

    1. An eventstream
    2. A stored procedure in a warehouse
    3. A Spark job definition Correct answer.
    4. A dataflow Gen2

    Explanation

    A Spark job definition runs a script against a lakehouse on demand or on a schedule, for example when invoked from a pipeline. It is intended for batch Spark workloads that do not need an interactive session.

  42. Question 42Implement and manage an analytics solution

    What does high concurrency mode provide for Fabric notebooks?

    1. Notebooks execute only on the capacity's driver node.
    2. Notebooks bypass the lakehouse and read data through the SQL analytics endpoint.
    3. Each user receives a dedicated Spark cluster that stays running indefinitely.
    4. Multiple users share a single Spark session, which reduces startup time and resource consumption. Correct answer.

    Explanation

    High concurrency mode lets several users or jobs share one Spark session in the same workspace, so sessions are reused instead of started per user. This lowers startup latency and resource use but introduces shared-session considerations for isolation.

  43. Question 43Implement and manage an analytics solution

    A pipeline must run the same copy activity once for each of 30 table names supplied as an array. Which activity provides the iteration?

    1. Switch
    2. Wait
    3. Lookup
    4. ForEach Correct answer.

    Explanation

    The ForEach activity iterates over a collection and runs the contained activities once per item. The collection can come from an array parameter, a variable, or the output of a Lookup activity.

  44. Question 44Implement and manage an analytics solution

    A pipeline must notify a Teams channel when a critical activity fails. What is the recommended configuration?

    1. Enable the secure output option on the activities.
    2. Set the pipeline concurrency to 1.
    3. Add a retry count of 3 to each activity.
    4. Connect the failure output of the critical activity to a Teams notification activity. Correct answer.

    Explanation

    Pipeline activities expose dependency outputs, including a failure path, so a notification activity can be chained to run only when the upstream activity fails. Retrying may recover transient errors but does not inform anyone when the failure persists.

  45. Question 45Implement and manage an analytics solution

    A notebook must read from an ADLS Gen2 account whose firewall permits only selected networks, and the account must not be opened to public access. Which Fabric capability allows this?

    1. Publishing the lakehouse to a Power BI workspace.
    2. Storing a shared access signature in a notebook cell.
    3. Enabling V-Order on the lakehouse tables.
    4. Configuring trusted workspace access for the workspace. Correct answer.

    Explanation

    Trusted workspace access allows a Fabric workspace to reach a firewall-protected storage account without exposing it to all networks, using the workspace identity for authorization. Network rules on the storage account are then satisfied by the trusted workspace.

  46. Question 46Implement and manage an analytics solution

    A notebook writes a large Delta table that is almost always filtered on a TransDate column. Which write option creates a physical layout that lets queries skip irrelevant files?

    1. Setting mergeSchema to true during the write.
    2. Setting overwriteSchema to true during the write.
    3. Creating a log checkpoint after every ten commits.
    4. Partitioning the table by the TransDate column during the write. Correct answer.

    Explanation

    Partitioning stores rows in separate folders keyed by the column value, so a filter on that column reads only the relevant folders. Schema options and checkpoints affect table evolution and log handling, not data pruning.

  47. Question 47Implement and manage an analytics solution

    In which format does a Fabric lakehouse store its tables by default?

    1. Avro
    2. CSV
    3. Delta Parquet Correct answer.
    4. JSON

    Explanation

    Lakehouse tables are Delta Lake tables, meaning Parquet data files plus a transaction log. This format enables time travel, ACID guarantees, and Direct Lake reading by semantic models.

  48. Question 48Implement and manage an analytics solution

    What must a workspace be assigned to before the Fabric items it contains can run?

    1. A separate compute cluster for each item.
    2. A Fabric capacity, such as an F SKU, or a Fabric trial capacity. Correct answer.
    3. A Power BI Pro per-user license.
    4. An Azure subscription that includes a storage account.

    Explanation

    Fabric workloads consume compute from a capacity, so a workspace must be assigned to a Fabric capacity or a trial capacity. Per-user Power BI licenses govern access to Power BI experiences but do not supply Fabric compute.

  49. Question 49Implement and manage an analytics solution

    A team must ingest streaming telemetry, query it with KQL, and retain a year of history. Which Fabric item should they use?

    1. A mirrored database
    2. A lakehouse
    3. An eventhouse that contains a KQL database Correct answer.
    4. A warehouse

    Explanation

    An eventhouse hosts KQL databases that are optimized for high-volume ingestion and KQL queries over telemetry. Retention policies on the KQL database control how long data is kept.

  50. Question 50Implement and manage an analytics solution

    A developer needs Delta time travel on a table to be available for 60 days. Which table property should be increased?

    1. delta.minReaderVersion
    2. delta.logRetentionDuration Correct answer.
    3. delta.columnMapping.mode
    4. delta.appendOnly

    Explanation

    The log retention duration controls how long historical versions remain in the transaction log, and time travel depends on those versions. The files must also be retained, so VACUUM must not remove data older than the desired window.

  51. Question 51Implement and manage an analytics solution

    A user must be able to create and edit items in a workspace but must not be able to add or remove workspace members. Which workspace role should be assigned to the user?

    1. Admin
    2. Viewer
    3. Member
    4. Contributor Correct answer.

    Explanation

    Contributor can create, edit, and delete workspace items but cannot add or remove members or change workspace settings. Member and Admin both include membership management, and Viewer is limited to read-only access.

  52. Question 52Implement and manage an analytics solution

    A team wants the definitions of its Fabric items stored as files in an Azure DevOps repository so that changes can be reviewed, branched, and rolled back. Which feature should be used?

    1. OneLake shortcuts
    2. Deployment pipelines
    3. Git integration Correct answer.
    4. Workspace identity

    Explanation

    Git integration connects a workspace to an Azure DevOps or GitHub repository and synchronizes item definitions as files, providing version history, branching, and pull-request review. Deployment pipelines move content between environments but do not store item definitions in a repository.

  53. Question 53Implement and manage an analytics solution

    A lakehouse must expose data that already resides in an external ADLS Gen2 account, and the data must not be duplicated into OneLake. What should be created?

    1. An external table
    2. A OneLake shortcut Correct answer.
    3. A mirrored database
    4. A data pipeline with a copy activity

    Explanation

    A OneLake shortcut is an object that points to data in another location, including ADLS Gen2, so queries read the files in place without copying them. Mirroring and copy activities both physically replicate data into OneLake.

  54. Question 54Implement and manage an analytics solution

    An engineer must review the status, duration, and history of pipeline, notebook, and dataflow runs across a workspace. Which Fabric feature should be used?

    1. The admin portal
    2. The capacity metrics app
    3. The workspace lineage view
    4. The Monitoring hub Correct answer.

    Explanation

    The Monitoring hub aggregates run history for pipelines, notebooks, dataflows, and refresh operations, showing status, start time, duration, and error details. Lineage shows data flow relationships, and the capacity metrics app reports on capacity consumption rather than individual item runs.

  55. Question 55Implement and manage an analytics solution

    A team needs near real-time replication of an Azure SQL database into Fabric as Delta tables, without authoring pipelines or notebooks. Which capability should they use?

    1. Dataflow Gen2
    2. Eventstream
    3. Database mirroring Correct answer.
    4. Deployment pipeline

    Explanation

    Mirroring continuously replicates a supported source database into OneLake as Delta tables with near real-time latency and requires no pipeline authoring. Dataflows run scheduled Power Query transformations, and eventstreams handle streaming event data rather than database replication.

  56. Question 56Implement and manage an analytics solution

    A semantic model must query very large Delta tables in a lakehouse with performance close to import mode, and the team does not want a scheduled import copy of the data. Which storage mode should be used?

    1. Direct Lake Correct answer.
    2. Import
    3. DirectQuery
    4. Dual

    Explanation

    Direct Lake reads Delta files directly from OneLake and loads column segments into memory on demand, giving near import-mode performance without a scheduled copy. DirectQuery sends each query to the source and is typically slower for large aggregations.

  57. Question 57Implement and manage an analytics solution

    A deployment pipeline promotes a semantic model from test to production, but the production model must connect to a different SQL server than the test model. What should be configured in the deployment pipeline?

    1. Parameter rules
    2. Deployment rules Correct answer.
    3. Stage filters
    4. Autobind policies

    Explanation

    Deployment rules let you override specific item properties, such as a data source connection or a parameter value, for a target stage so the same item definition works across environments. Without a rule, the item keeps the value from the source stage.

  58. Question 58Implement and manage an analytics solution

    A business analyst must apply Power Query transformations without writing code and land the results in a lakehouse on a recurring schedule. Which Fabric item should be used?

    1. A dataflow Gen2 Correct answer.
    2. A mirrored database
    3. A notebook
    4. A stored procedure

    Explanation

    Dataflow Gen2 provides the Power Query Online experience for low-code transformations and can write results to a lakehouse or warehouse with a scheduled refresh. Notebooks require code, and mirroring replicates databases without transformation logic.

  59. Question 59Implement and manage an analytics solution

    A shortcut to ADLS Gen2 must be authorized so that scheduled jobs can read the data unattended, without depending on any individual user's credentials. What should be used to authorize the connection?

    1. A stored credential in each notebook
    2. The capacity admin account
    3. The item owner's user token
    4. The workspace identity Correct answer.

    Explanation

    A workspace identity is a managed identity associated with the workspace that items can use to authenticate to data sources, so unattended jobs do not rely on a person's credentials and are not broken when that person leaves. Storing credentials in code creates maintenance and security problems.

  60. Question 60Implement and manage an analytics solution

    A pipeline must process data in fixed, non-overlapping time periods, retry failed periods, and backfill historical periods on demand. Which trigger type should be used?

    1. Event-based trigger
    2. Schedule trigger
    3. Tumbling window Correct answer.
    4. Manual trigger

    Explanation

    A tumbling window trigger runs for fixed, non-overlapping time windows, keeps state per window, supports retries, and can backfill past windows. A schedule trigger fires at points in time without window state or backfill semantics.

  61. Question 61Implement and manage an analytics solution

    A data engineer must load millions of rows from Parquet files into a Fabric warehouse table as quickly as possible. Which T-SQL approach should be used?

    1. INSERT ... SELECT from the lakehouse SQL endpoint
    2. An UPDATE statement with a JOIN
    3. A stored procedure that loops over rows
    4. The COPY INTO statement Correct answer.

    Explanation

    COPY INTO performs a high-throughput bulk load directly from files in a lakehouse or external storage into a warehouse table. Row-by-row statements and INSERT ... SELECT are far slower for large file-based loads.

  62. Question 62Implement and manage an analytics solution

    A notebook must invoke a second notebook, pass parameters to it, and use the value that the second notebook returns. Which approach should be used?

    1. A %pip install command in the first cell
    2. dbutils.fs.cp to stage the second notebook
    3. An external spark-submit job definition
    4. mssparkutils.notebook.run Correct answer.

    Explanation

    mssparkutils.notebook.run executes another notebook as a child activity with parameters and returns its exit value to the parent notebook. The %run magic executes a notebook inline without returning a value or accepting the same parameter passing.

  63. Question 63Implement and manage an analytics solution

    A compliance team wants a sensitivity label applied once so that reports built on the data inherit the same protection. Where should the label be applied?

    1. On the workspace
    2. On the capacity
    3. On each report separately
    4. On the source lakehouse Correct answer.

    Explanation

    Sensitivity labels applied to Fabric data items such as lakehouses and warehouses propagate to downstream items that consume that data, so reports inherit the label. Applying labels per report is error-prone and does not cover all consumers.

  64. Question 64Implement and manage an analytics solution

    Which endorsement can only be granted by a Fabric administrator or a domain administrator and indicates that data meets organizational quality standards?

    1. A workspace role
    2. Certification Correct answer.
    3. Promotion
    4. A sensitivity label

    Explanation

    Certification is a restricted endorsement that only authorized reviewers such as Fabric admins or domain admins can grant, signaling that the item meets organizational standards. Promotion is a lighter endorsement that item owners with write permission can apply themselves.

  65. Question 65Implement and manage an analytics solution

    A Direct Lake semantic model queries a Delta table that contains a column whose data type is not supported by Direct Lake, and a report requests that column. What happens?

    1. The table is converted to Import mode automatically
    2. The column is silently omitted from the results
    3. The model falls back to DirectQuery for that query Correct answer.
    4. The query fails until the column is removed

    Explanation

    When Direct Lake encounters a feature it cannot serve directly, such as an unsupported data type in a referenced column, the model falls back to DirectQuery so the query still returns results, at the cost of performance. The table is not converted to Import mode and results are not silently altered.

  66. Question 66Implement and manage an analytics solution

    Users report that interactive queries in a workspace slow down sharply while a large Spark job runs, and both items are assigned to the same small Fabric capacity. What is the most likely cause?

    1. The semantic model automatically switched to DirectQuery for the duration of the job.
    2. The Spark job has invalidated the OneLake shortcuts used by the interactive queries.
    3. The workspace was temporarily moved to a trial capacity while the Spark job was running.
    4. Interactive requests are being throttled by capacity smoothing. Correct answer.

    Explanation

    Fabric capacities smooth consumption over time, and when background operations consume a large share of the capacity, interactive workloads are throttled and experience delays until utilization drops. Recovery usually involves waiting, rescheduling the job, or scaling the capacity.

  67. Question 67Ingest and transform data

    Which Fabric item provides a low-code, Power Query-based experience for building data transformations and loading the results into a Lakehouse or Warehouse?

    1. Dataflow Gen2 Correct answer.
    2. Notebook
    3. Eventstream
    4. Data pipeline

    Explanation

    Dataflow Gen2 uses the Power Query Online editor, so transformations are authored visually with a low-code approach, and the query results can be written to Fabric destinations such as a Lakehouse or Warehouse. Eventstream, notebooks, and pipelines serve different purposes (streaming, code-first processing, and orchestration) rather than visual Power Query authoring.

  68. Question 68Ingest and transform data

    A pipeline must execute a Fabric notebook as one step in a multi-step workflow. Which pipeline activity performs this?

    1. Web activity
    2. Stored procedure activity
    3. Notebook activity Correct answer.
    4. Script activity

    Explanation

    The Notebook activity in a Fabric data pipeline runs a referenced notebook and can receive parameters, making it the correct choice for executing notebook code as a pipeline step. The Script, Stored procedure, and Web activities target other execution surfaces.

  69. Question 69Ingest and transform data

    What does a OneLake shortcut provide?

    1. A cached materialized view of the source that is refreshed on an hourly schedule
    2. A scheduled copy of the data into the Lakehouse that refreshes on a timer
    3. A virtual reference that reads the external data in place Correct answer.
    4. A fully mirrored replica of the source kept continuously in sync

    Explanation

    A shortcut is an object in OneLake that points to data stored elsewhere, such as ADLS Gen2, Amazon S3, or another Fabric location, and reads it without copying it. Because no data is duplicated, the source remains the single copy.

  70. Question 70Ingest and transform data

    Which Fabric item ingests streaming events and routes them to one or more destinations in near real time?

    1. Eventhouse
    2. Eventstream Correct answer.
    3. Data pipeline
    4. Dataflow Gen2

    Explanation

    An Eventstream is the streaming ingestion item: events are captured from sources such as Event Hubs or Kafka and routed to destinations like a Lakehouse, Eventhouse, or Activator. Eventhouse stores and queries the data rather than performing the routing.

  71. Question 71Ingest and transform data

    Which data source can be continuously replicated into Fabric by using database mirroring?

    1. OneLake
    2. Azure Blob Storage
    3. Azure Cosmos DB Correct answer.
    4. SharePoint Online

    Explanation

    Mirroring continuously replicates a source database into OneLake as Delta tables, and Azure Cosmos DB is one of the supported mirrored sources, along with sources such as Azure SQL Database and Snowflake. Storage services and file repositories are accessed through shortcuts instead of mirroring.

  72. Question 72Ingest and transform data

    Which storage format underlies the tables created in a Fabric Lakehouse?

    1. ORC
    2. Avro
    3. Delta Correct answer.
    4. CSV

    Explanation

    Tables in a Lakehouse are Delta tables, meaning Parquet data files plus a Delta transaction log that provides ACID transactions and schema enforcement. CSV, Avro, and ORC can exist as files in the Files area but are not the table format used for managed Lakehouse tables.

  73. Question 73Ingest and transform data

    Which T-SQL statement loads data from files stored in OneLake into a table in a Fabric Warehouse?

    1. MERGE
    2. COPY INTO Correct answer.
    3. INSERT ... SELECT from OPENROWSET
    4. BULK INSERT

    Explanation

    COPY INTO is the T-SQL statement supported by the Fabric Warehouse for high-throughput ingestion from files such as Parquet in OneLake or Azure storage. It handles file parsing and loading in a single statement.

  74. Question 74Ingest and transform data

    In a medallion architecture, which layer stores raw data exactly as it was ingested, with no cleansing applied?

    1. Gold
    2. Bronze Correct answer.
    3. Semantic
    4. Silver

    Explanation

    The bronze layer holds raw ingested data in its original shape, preserving history for reprocessing. Silver contains cleansed and conformed data, and gold contains business-level aggregates and curated marts.

  75. Question 75Ingest and transform data

    A ForEach activity loops over 40 items and currently processes them one at a time. Which change makes the iterations run concurrently?

    1. Increase the Copy activity retry count
    2. Clear the Sequential option Correct answer.
    3. Enable the Wait activity before each loop
    4. Set the batch count property to zero

    Explanation

    The ForEach activity has a Sequential option that is enabled by default, forcing iterations to run one after another. Clearing it lets iterations run in parallel, with the batch count controlling how many run at once.

  76. Question 76Ingest and transform data

    Which dynamic content expression returns the current UTC timestamp when used in a pipeline activity?

    1. @currentTime()
    2. @timestamp()
    3. @utcNow() Correct answer.
    4. @guid()

    Explanation

    The @utcNow() function returns the current date and time in UTC and is commonly used to build date-based folder paths or file names. Helper functions such as @guid() and string or date conversion functions serve different purposes.

  77. Question 77Ingest and transform data

    A DataFrame has a column containing arrays. Which PySpark function produces one output row for each element of that array?

    1. flatten
    2. pivot
    3. explode Correct answer.
    4. collect_list

    Explanation

    explode() takes an array column and returns a new row for every element, which is the standard way to flatten nested arrays into rows. collect_list performs the opposite aggregation, grouping values into an array.

  78. Question 78Ingest and transform data

    Which credentials does a OneLake shortcut to an Amazon S3 bucket require?

    1. A shared access signature token
    2. An OAuth client credential
    3. An access key and secret key Correct answer.
    4. A Microsoft Entra service principal

    Explanation

    The Amazon S3 shortcut connector authenticates with an S3 access key ID and secret access key, which are stored as the shortcut's connection credentials. Shared access signatures and Entra credentials apply to Azure storage and Azure services instead.

  79. Question 79Ingest and transform data

    A pipeline loads data incrementally using a watermark value. What is the purpose of the Lookup activity placed before the Copy activity?

    1. It retrieves the last stored watermark value Correct answer.
    2. It starts the schedule trigger for the next pipeline run
    3. It truncates the destination table before the copy starts
    4. It validates that the source file schema matches the target

    Explanation

    In a watermark pattern the Lookup activity reads the previously recorded high-water mark from a control table, and the Copy activity uses that value in its filter to fetch only newer rows. After a successful load, another activity writes the new watermark back.

  80. Question 80Ingest and transform data

    Which Spark SQL command compacts a Delta table by rewriting many small files into a smaller number of larger files?

    1. VACUUM
    2. CACHE TABLE
    3. OPTIMIZE Correct answer.
    4. ANALYZE TABLE

    Explanation

    OPTIMIZE performs file compaction on a Delta table by bin-packing small files into larger ones, which improves read performance for subsequent queries. In Fabric the rewritten files also receive V-Order processing.

  81. Question 81Ingest and transform data

    A pipeline must start as soon as a new file is created in an ADLS Gen2 account. Which trigger type should be configured?

    1. Schedule trigger
    2. Manual trigger
    3. Event trigger Correct answer.
    4. Tumbling window trigger

    Explanation

    An event trigger subscribes to storage events and starts the pipeline when a matching event such as a created blob occurs, which removes polling delays. A schedule trigger only runs at fixed times regardless of file arrival.

  82. Question 82Ingest and transform data

    An Eventhouse table must be readable from OneLake and other Fabric engines without duplicating the data. Which feature should be enabled?

    1. A OneLake shortcut to the KQL tables
    2. Continuous export to a storage account
    3. OneLake availability Correct answer.
    4. Database mirroring of the KQL database

    Explanation

    Enabling OneLake availability on an Eventhouse database makes its tables available in OneLake as Delta-formatted data that other Fabric engines can read, without a separate copy pipeline. Continuous export and shortcuts are different mechanisms that involve additional copies or references.

  83. Question 83Ingest and transform data

    After a load completes, logic implemented as a stored procedure in a Fabric Warehouse must run. Which pipeline activity should be added?

    1. Web activity that calls the REST endpoint
    2. Stored procedure activity Correct answer.
    3. Copy activity with a binary dataset
    4. Lookup activity that runs a query

    Explanation

    The Stored procedure activity invokes a stored procedure in a supported database, including a Fabric Warehouse, and can pass parameters from the pipeline. It is the purpose-built activity for executing database-side logic as a pipeline step.

  84. Question 84Ingest and transform data

    A pipeline copies data from an on-premises SQL Server instance. What must be deployed and configured to reach that source?

    1. An ExpressRoute circuit to the Azure region
    2. A managed private endpoint in the workspace
    3. An on-premises data gateway Correct answer.
    4. A private link service in the virtual network

    Explanation

    Data that lives on-premises is reached through an on-premises data gateway, which brokers the connection between Fabric and the local network. Managed private endpoints and private link services handle private connectivity to Azure resources, not to a local SQL Server.

  85. Question 85Ingest and transform data

    How does the Tables area of a Lakehouse differ from the Files area?

    1. Tables contains Delta tables surfaced by the SQL analytics endpoint Correct answer.
    2. Tables is read-only from Spark, whereas Files can be written by notebooks and pipelines
    3. Files can hold only CSV files, whereas Tables can hold any file format supported by Spark
    4. Files requires an assigned schema, whereas Tables are always schema-less containers

    Explanation

    Anything written to the Tables area as a Delta table is registered in the Lakehouse metastore and becomes queryable through the SQL analytics endpoint and the default semantic model. The Files area stores arbitrary files that have no automatic table registration.

  86. Question 86Ingest and transform data

    In a notebook you must insert new rows and update existing rows in a Delta table by matching on a key column. Which operation should you use?

    1. INSERT OVERWRITE
    2. COPY INTO
    3. VACUUM
    4. MERGE INTO Correct answer.

    Explanation

    MERGE INTO compares the source rows to the target using a match condition and applies the requested insert, update, or delete actions, which is the standard upsert pattern. INSERT OVERWRITE replaces all data and COPY INTO only appends.

  87. Question 87Ingest and transform data

    Which statement about V-Order in Fabric is accurate?

    1. It is applied at read time by the query engine and has no impact on how the data is stored on disk
    2. It must be enabled manually for every table because Fabric disables it by default at the workspace level
    3. It is applied at write time and speeds up reads at the cost of slower writes Correct answer.
    4. It removes the Delta transaction log so that readers must infer the schema from the Parquet files

    Explanation

    V-Order is a write-time optimization that reorganizes Parquet data, applying sorting, dictionary encoding, and compression so that read-optimized engines such as the SQL analytics endpoint and Power BI can scan it faster. The trade-off is additional work during writes.

  88. Question 88Ingest and transform data

    A pipeline must load each source file exactly once, even when the pipeline is rerun after a failure. Which design achieves this?

    1. Turn on the Sequential setting on the ForEach activity so iterations cannot overlap
    2. Set the Copy activity retry count to zero so failed slices are never re-run by the pipeline
    3. Keep a control table of processed file names and skip files already recorded Correct answer.
    4. Attach a schedule trigger that runs every five minutes and rely on the copy to deduplicate

    Explanation

    Idempotency requires state outside the pipeline. Recording each successfully loaded file in a control table and checking that table before copying ensures a rerun skips files that already landed, regardless of how many times the pipeline executes.

  89. Question 89Ingest and transform data

    What does enabling caching on a OneLake shortcut do?

    1. It copies the entire external dataset permanently into the Lakehouse as Delta tables
    2. It compresses the files at the source system to reduce network transfer
    3. It converts the external files into managed Delta tables on first access
    4. It stores recently read files in a local cache that expires Correct answer.

    Explanation

    Shortcut caching keeps a copy of files that have been accessed in a Fabric-managed cache close to compute, so repeated reads avoid going back to the remote source. Cached entries expire after the configured period, at which point the data is read from the source again.

  90. Question 90Ingest and transform data

    Which Eventstream destination writes streaming events as Delta tables that can be queried through a Lakehouse SQL analytics endpoint?

    1. Lakehouse Correct answer.
    2. Eventhouse
    3. Custom endpoint
    4. Activator

    Explanation

    A Lakehouse destination in an Eventstream writes the incoming events into a Delta table in the attached Lakehouse, which then becomes available through the SQL analytics endpoint and the semantic model. Eventhouse destinations land the data in a KQL database instead.

  91. Question 91Ingest and transform data

    Inside a ForEach activity, which expression references the value of the current iteration when it is used in a nested Copy activity?

    1. @item() Correct answer.
    2. @variables('fileName')
    3. @pipeline().parameters.fileName
    4. @activity('Lookup1').output

    Explanation

    Within a ForEach activity, @item() resolves to the current element of the collection being iterated, so each iteration can build a distinct source path or table name. Pipeline parameters stay constant for the whole run, and activity outputs reference a specific earlier activity result.

  92. Question 92Ingest and transform data

    A Dataflow Gen2 is used to cleanse data and then publish the result. Which item can be configured as a data destination for the dataflow?

    1. An Eventstream
    2. A KQL queryset
    3. A Real-Time Dashboard
    4. A Lakehouse table Correct answer.

    Explanation

    Dataflow Gen2 writes its output to supported data destinations such as Lakehouse tables, Warehouse tables, and Azure SQL tables. Streaming and visualization items are not data destinations.

  93. Question 93Ingest and transform data

    A Copy activity reads CSV files in which some rows contain values that cannot be converted to the sink column types. You want the load to continue and the bad rows to be recorded. Which Copy activity setting should you enable?

    1. Staging
    2. Schema drift
    3. Parallel copy
    4. Fault tolerance Correct answer.

    Explanation

    Fault tolerance makes the Copy activity skip incompatible rows and log them to a store you configure, so the rest of the load can continue.

  94. Question 94Ingest and transform data

    Which KQL command appends the results of a query to an existing table in an Eventhouse database?

    1. .show tables
    2. .create table
    3. .set-or-append Correct answer.
    4. .drop table

    Explanation

    .set-or-append writes query results into a table, creating the table if needed and appending when it already exists. The other commands change metadata or list objects.

  95. Question 95Ingest and transform data

    In a Fabric notebook, which magic command at the start of a cell makes that cell execute as T-SQL?

    1. %%markdown
    2. %%pyspark
    3. %%configure
    4. %%sql Correct answer.

    Explanation

    Language magics switch the interpreter for a single cell; %%sql runs the cell against the SQL endpoint of the attached Lakehouse or Warehouse.

  96. Question 96Ingest and transform data

    Activity B must run after Activity A finishes, whether A succeeds or fails. Which dependency condition should be set for Activity B?

    1. Skipped
    2. Completed Correct answer.
    3. Succeeded
    4. Failed

    Explanation

    The Completed condition fires when the upstream activity reaches any terminal state, including success and failure.

  97. Question 97Ingest and transform data

    Which Fabric capability continuously replicates an Azure SQL Database into OneLake with no pipeline authoring?

    1. Shortcuts
    2. Dataflow Gen2
    3. Mirroring Correct answer.
    4. Eventstream

    Explanation

    Mirroring continuously replicates a supported source database into OneLake in near real time so Fabric engines can query the replica.

  98. Question 98Ingest and transform data

    When schemas are enabled on a Lakehouse, how are the tables organized?

    1. Into one schema per workspace
    2. As folders under Files
    3. In one flat namespace
    4. Into schemas that group tables Correct answer.

    Explanation

    Schema-enabled Lakehouses group tables into schemas, with dbo as the default, so names can be qualified as schema.table.

  99. Question 99Ingest and transform data

    Which Delta Lake operation removes data files that are no longer referenced and are older than the retention interval?

    1. VACUUM Correct answer.
    2. FSCK REPAIR
    3. ZORDER
    4. OPTIMIZE

    Explanation

    VACUUM deletes unreferenced files past the retention threshold, after which time travel to versions older than that threshold stops working.

  100. Question 100Ingest and transform data

    A Copy activity loads a large table from an on-premises Oracle database into a Fabric Warehouse. Which feature is used to land the data before it is inserted into the warehouse table?

    1. Fault tolerance
    2. Schema drift
    3. Data preview
    4. Staging Correct answer.

    Explanation

    For warehouse sinks, the Copy activity can stage extracted data in OneLake or ADLS before the final load, which is the recommended pattern for large volumes.

  101. Question 101Ingest and transform data

    A pipeline must refresh a Dataflow Gen2 as one step in a sequence. Which pipeline activity invokes the dataflow?

    1. Copy
    2. Lookup
    3. Script
    4. Dataflow Correct answer.

    Explanation

    The Dataflow activity runs a Dataflow Gen2 and can be sequenced with other pipeline activities.

  102. Question 102Ingest and transform data

    A Copy activity reads Parquet files whose columns change between batches. New source columns must be added to the sink table automatically. Which setting should you enable?

    1. Enable staging
    2. Allow schema drift Correct answer.
    3. Preserve hierarchy
    4. Fault tolerance

    Explanation

    Schema drift lets the Copy activity pass through source columns that are not defined in the sink schema, adding them when the sink supports it.

  103. Question 103Ingest and transform data

    A Spark notebook writes a DataFrame to a Delta table that may already contain rows. Existing rows must be kept. Which save mode should be used?

    1. overwrite
    2. error
    3. append Correct answer.
    4. ignore

    Explanation

    Append adds the DataFrame rows to the existing table while leaving current data in place, whereas overwrite replaces the table contents.

  104. Question 104Ingest and transform data

    Which pipeline activity assigns a value to a pipeline variable so later activities can read it with @variables('name')?

    1. Set Variable Correct answer.
    2. Copy Data activity
    3. Lookup activity
    4. Wait activity

    Explanation

    Set Variable updates a pipeline variable at run time, and downstream activities read it through the variables() expression.

  105. Question 105Ingest and transform data

    A pipeline processes fixed historical time slices and must support rerunning a specific past window with its own state and retries. Which trigger type should be used?

    1. Tumbling window Correct answer.
    2. Event trigger
    3. Manual trigger
    4. Schedule trigger

    Explanation

    Tumbling window triggers keep per-window state, run non-overlapping time slices, and support backfilling past windows.

  106. Question 106Ingest and transform data

    Which Eventhouse feature automatically runs a query over newly ingested records in a source table and appends the results to a target table?

    1. Update policy Correct answer.
    2. Stored function
    3. Continuous export
    4. Materialized view

    Explanation

    An update policy is attached to a target table and runs a query over newly ingested source records, appending the results to that table.

  107. Question 107Ingest and transform data

    Which notebook magic command installs a Python library for use in the current Spark session?

    1. %%configure
    2. %%sql
    3. %run
    4. %pip Correct answer.

    Explanation

    %pip installs Python packages into the current notebook session; the other magics change language, configuration, or run another notebook.

  108. Question 108Ingest and transform data

    A Copy activity reads one very large table and must increase throughput by issuing multiple read requests in parallel. Which setting controls this?

    1. Allow schema drift
    2. Enable fault tolerance
    3. Enable staging
    4. Set copy parallelism Correct answer.

    Explanation

    The degree of copy parallelism (parallelCopies) sets how many concurrent read and write threads a single Copy activity uses.

  109. Question 109Ingest and transform data

    In Eventstream, which operator creates a new stream from an existing stream so that different processing can feed different destinations?

    1. Destination
    2. Source
    3. Derived stream Correct answer.
    4. Window

    Explanation

    A derived stream branches from an existing stream, allowing separate transformations and destinations to consume the same source data.

  110. Question 110Ingest and transform data

    A Dataflow Gen2 loads a large table and should retrieve only rows that changed since the last refresh. Which capability should be configured?

    1. Staging
    2. Data destination configuration
    3. Incremental refresh Correct answer.
    4. Query folding

    Explanation

    Incremental refresh stores a reference set of historical data and refreshes only rows that fall inside the configured refresh period.

  111. Question 111Ingest and transform data

    A pipeline must pass a date value into a notebook so the notebook can filter data. Which approach accomplishes this?

    1. Store the date in a Spark configuration setting
    2. Use a notebook parameter supplied by the pipeline Correct answer.
    3. Add a Wait activity before the notebook
    4. Hard-code the date inside a notebook code cell

    Explanation

    Parameters declared in a notebook are exposed as base parameters on the Notebook activity, and the pipeline supplies their values at run time.

  112. Question 112Ingest and transform data

    Two Spark jobs attempt to commit writes to the same Delta table at nearly the same time. Which mechanism keeps the commits consistent?

    1. File locks issued by OneLake
    2. VACUUM with short retention
    3. Z-ordering on the key column
    4. Optimistic concurrency Correct answer.

    Explanation

    Delta Lake records each commit in a transaction log and uses optimistic concurrency, so a conflicting writer retries or fails according to isolation rules.

  113. Question 113Ingest and transform data

    A Delta table has been modified several times. Which statement returns the table as it existed after the third commit, without changing the current table?

    1. SELECT * FROM t TIMESTAMP AS OF 3
    2. VACUUM t RETAIN 3 HOURS
    3. RESTORE TABLE t TO VERSION 3
    4. SELECT * FROM t VERSION AS OF 3 Correct answer.

    Explanation

    The VERSION AS OF clause reads a historical snapshot identified by its commit version while leaving the current table unchanged.

  114. Question 114Ingest and transform data

    A Dataflow Gen2 query reads from a SQL source and includes a step that calls a custom function not supported by the source. What is the effect on query folding?

    1. Folding stops there; later steps run locally Correct answer.
    2. Folding continues and the step becomes SQL
    3. The dataflow fails when it refreshes
    4. The dataflow switches to DirectQuery mode

    Explanation

    Query folding pushes transformations to the source until a step cannot be translated; that step and everything after it execute in the mashup engine.

  115. Question 115Ingest and transform data

    A Spark job must replace only the rows in a Delta table that match a date predicate and leave all other rows unchanged. Which write option should be used?

    1. replaceWhere Correct answer.
    2. mergeSchema
    3. mode('overwrite')
    4. mode('append')

    Explanation

    replaceWhere performs an atomic conditional overwrite, so only rows matching the predicate are replaced and the rest of the table is untouched.

  116. Question 116Ingest and transform data

    Which trade-off applies when streaming ingestion is enabled on a KQL database?

    1. Longer data retention with no effect on latency
    2. Lower latency but lower throughput than queued Correct answer.
    3. Removal of the need to define a table schema
    4. Automatic conversion of tables to external tables

    Explanation

    Streaming ingestion is designed for low-latency, modest-volume pipelines, whereas queued ingestion batches data and scales to higher throughput.

  117. Question 117Ingest and transform data

    A team has Parquet files in an Azure Data Lake Storage Gen2 container. They create a Lakehouse shortcut that points at the container. What happens to the data?

    1. The files are copied once into the lakehouse Files area when the shortcut is created.
    2. The files are converted into a Delta table that the shortcut exposes.
    3. The files are read from ADLS Gen2 on demand; no copy of the data is made. Correct answer.
    4. The files are cached in OneLake and refreshed on a nightly schedule.

    Explanation

    A shortcut is a reference to data that stays in its original location. OneLake stores only the pointer, so queries read the source files directly and no duplicate copy is created or synchronized.

  118. Question 118Ingest and transform data

    A Copy activity loads CSV files in which a small number of rows contain values that cannot be converted to the sink column types. The load must continue and the rejected rows must be recorded. What should you configure?

    1. Enable staging on the copy activity's linked service.
    2. Increase the data integration units allocated to the copy activity.
    3. Configure fault tolerance to skip incompatible rows and write them to a log. Correct answer.
    4. Set the write batch size of the copy activity to zero.

    Explanation

    The fault-tolerance settings on a Copy activity control whether rows that fail conversion are skipped and where they are logged. Enabling it lets the rest of the load complete instead of failing the whole activity.

  119. Question 119Ingest and transform data

    In a Fabric Warehouse, which statement both creates a new table and populates it from a query in a single operation?

    1. SELECT * INTO dbo.OpenOrders FROM dbo.Orders WHERE Status = 'Open';
    2. CREATE TABLE dbo.OpenOrders AS SELECT * FROM dbo.Orders WHERE Status = 'Open'; Correct answer.
    3. INSERT INTO dbo.OpenOrders SELECT * FROM dbo.Orders WHERE Status = 'Open';
    4. ALTER TABLE dbo.Orders SWITCH TO dbo.OpenOrders;

    Explanation

    Fabric Warehouse supports CREATE TABLE AS SELECT (CTAS) to define a table and fill it from a query in one statement. SELECT INTO is not supported in the Warehouse, and INSERT requires the target table to exist already.

  120. Question 120Ingest and transform data

    A pipeline must execute a T-SQL stored procedure in a Fabric Warehouse after a load finishes. Which pipeline activity should be used?

    1. Web activity
    2. Copy activity
    3. Stored procedure activity Correct answer.
    4. Lookup activity

    Explanation

    The Stored procedure activity is designed to invoke a stored procedure on a database, including a Fabric Warehouse, as a pipeline step. Other activities have different purposes and do not take a procedure name and parameters.

  121. Question 121Ingest and transform data

    A Dataflow Gen2 query combines two sources, and the result must be persisted as a table in a Lakehouse for downstream Spark notebooks. What must be added to the dataflow?

    1. An output destination that targets the Lakehouse. Correct answer.
    2. A staging account for the dataflow.
    3. A gateway that connects to the Lakehouse.
    4. A compute scale setting on the dataflow.

    Explanation

    A Dataflow Gen2 persists results only when an output destination is defined. Pointing the destination at the Lakehouse writes the query result as a table that other items can read.

  122. Question 122Ingest and transform data

    A pipeline must run the same Copy activity once for every row returned by a query against a control table. Which pattern achieves this?

    1. A Get Metadata activity followed by a Switch activity.
    2. A Lookup activity followed by a ForEach activity that contains the Copy activity. Correct answer.
    3. An Until activity that contains a Copy activity and a Set variable activity.
    4. A Filter activity followed by a Wait activity.

    Explanation

    A Lookup activity returns the query result as an array, and a ForEach activity iterates over that array so the inner Copy activity runs once per row with @item() available.

  123. Question 123Ingest and transform data

    A folder in a lakehouse Files area receives a new CSV file each hour. A pipeline must copy every file present in the folder into a Warehouse table. Which two activities form the pattern?

    1. Get Metadata activity and ForEach activity Correct answer.
    2. Copy activity and Script activity
    3. Wait activity and Set variable activity
    4. Lookup activity and Until activity

    Explanation

    Get Metadata returns the folder's childItems array, and ForEach iterates over that array so a Copy activity inside the loop processes each file. The file name is read with @item().name.

  124. Question 124Ingest and transform data

    A pipeline loads only rows whose ModifiedDate is later than the value loaded by the previous successful run, and that value must survive future runs. Where should the value be kept?

    1. In a pipeline variable with a default value defined on the pipeline.
    2. In the pipeline run identifier exposed by the service.
    3. In the pre-copy script property of the Copy activity.
    4. In a control table in a warehouse that the pipeline reads and updates. Correct answer.

    Explanation

    Pipeline variables are reinitialized on every run, so a durable high-water mark must live outside the pipeline. A control table is read at the start of the run and updated after a successful load.

  125. Question 125Ingest and transform data

    A notebook uses spark.readStream and writeStream to continuously append to a Delta table. Which setting must be supplied so the query tracks its progress and can recover after a restart?

    1. A checkpointLocation in the write stream options. Correct answer.
    2. The mergeSchema option set to true.
    3. A repartition of the source to a single partition.
    4. A trigger interval of zero.

    Explanation

    Structured streaming stores progress and state in a checkpoint location, which is required to recover a query after restart. Without it the query cannot reliably resume where it stopped.

  126. Question 126Ingest and transform data

    A Dataflow Gen2 writes a large result to a Fabric Warehouse, and refreshes are slow because transformed rows are inserted one at a time. Which feature should be enabled?

    1. Fast copy on the dataflow's destination.
    2. Staging, backed by a Lakehouse. Correct answer.
    3. Query folding on the source query.
    4. Incremental refresh on the dataflow.

    Explanation

    Staging writes the transformed output to a Lakehouse first and then performs a single bulk load into the warehouse. That bulk operation is far faster than inserting rows individually.

  127. Question 127Ingest and transform data

    A Delta table in a lakehouse receives many small appends, and read performance has degraded because of the large number of small files. Which command addresses this?

    1. Run OPTIMIZE on the table. Correct answer.
    2. Run VACUUM on the table.
    3. Run RESTORE on the table.
    4. Run DESCRIBE HISTORY on the table.

    Explanation

    OPTIMIZE compacts small files into larger ones, which reduces per-file overhead during reads. VACUUM removes unreferenced files but does not merge the small files that remain active.

  128. Question 128Ingest and transform data

    In Eventstream, you must calculate the average sensor reading per device over fixed five-minute intervals before sending the results to a destination. Which operator do you add?

    1. Union of two input streams
    2. Expand a nested record column
    3. Group by with a tumbling window Correct answer.
    4. Manage fields: drop unused columns

    Explanation

    Eventstream aggregation is performed by the Group by operator, which applies a window such as tumbling to compute values over time intervals. Grouping by device yields one result per device per window.

  129. Question 129Ingest and transform data

    A Fabric Warehouse must query a Delta table that resides in a Lakehouse in the same workspace, and the data must not be copied. What is the supported approach?

    1. Export the table to Parquet and read it with OPENROWSET.
    2. Create a mirrored database over the Lakehouse table.
    3. Query the table with a three-part name such as Lakehouse.dbo.TableName. Correct answer.
    4. Load the table into the Warehouse with a Copy activity first.

    Explanation

    Fabric supports cross-database queries within a workspace, so a warehouse can read lakehouse tables by referencing them with a three-part name. No data movement is involved, which satisfies the no-copy requirement.

  130. Question 130Ingest and transform data

    A notebook must run for the session on a larger driver node with more executors than the pool default. Which approach applies this for the whole session?

    1. Run %%configure in the first cell of the notebook. Correct answer.
    2. Install a resource package with %pip and restart the session.
    3. Pass the values as a pipeline variable to the notebook activity.
    4. Call spark.conf.set from the final cell of the notebook.

    Explanation

    %%configure recreates the Spark session with the specified driver and executor sizes and counts, and it must be the first command in the notebook. Settings applied later in the notebook cannot resize the running session.

  131. Question 131Ingest and transform data

    A Spark notebook needs to write a Parquet file into the Files area of the lakehouse it is attached to. Which path should the notebook use?

    1. C:/lakehouse/Files/out
    2. dbfs:/mnt/lakehouse/Files/out
    3. Files/out Correct answer.
    4. abfss://<workspace>@onelake.dfs.fabric.microsoft.com/<lakehouse>/Files/out

    Explanation

    A notebook attached to a lakehouse resolves relative paths against that default lakehouse. Using Files/out therefore targets the Files section without hardcoding workspace or item names.

  132. Question 132Ingest and transform data

    A Spark notebook joins a 500 GB fact table to a 20 MB dimension table, and the job spends most of its time in a shuffle exchange. Which change is most likely to remove that shuffle?

    1. Broadcast the small dimension table in the join. Correct answer.
    2. Repartition both tables to 1000 partitions before the join.
    3. Write both tables to Delta and read them back before joining.
    4. Cache the fact table in memory before the join.

    Explanation

    A broadcast join ships the small table to every executor, so the large table can be joined partition-locally without being redistributed. Because the dimension table is only 20 MB, broadcasting it is feasible and eliminates the shuffle of the fact table.

  133. Question 133Monitor and optimize an analytics solution

    Which Fabric feature provides a single, cross-workspace view of the run history for pipelines, notebooks, dataflows, and Spark jobs?

    1. The Monitoring hub Correct answer.
    2. The Real-Time hub
    3. The OneLake catalog
    4. The workspace lineage view

    Explanation

    The Monitoring hub aggregates activity across workspaces so you can track the status and duration of pipelines, notebooks, dataflows, and other item runs in one place. It is the entry point for drilling into a specific run to view its detail. The Real-Time hub and OneLake catalog serve different discovery purposes.

  134. Question 134Monitor and optimize an analytics solution

    Which tool should you use to analyze how much of a Fabric capacity is consumed by individual workspaces and items over time?

    1. The lineage view
    2. The Monitoring hub
    3. The Capacity Metrics app Correct answer.
    4. The Real-Time hub

    Explanation

    The Microsoft Fabric Capacity Metrics app is a Power BI template app that reports compute consumption, throttling, and overage broken down by workspace, item, and operation over a selected time window. The Monitoring hub tracks run status rather than consumption percentages.

  135. Question 135Monitor and optimize an analytics solution

    For a Spark notebook run, which interface shows stage- and task-level execution details such as shuffles and spills?

    1. The workspace lineage view
    2. The Monitoring hub run detail pane
    3. The Capacity Metrics app
    4. The Apache Spark UI Correct answer.

    Explanation

    The Spark UI exposes jobs, stages, tasks, shuffle and spill metrics, and executor logs for a specific Spark application, which is what you need for low-level performance tuning. The Monitoring hub provides run-level status and links to that deeper view.

  136. Question 136Monitor and optimize an analytics solution

    Where can you review the previous refresh outcomes and durations for a semantic model?

    1. In the semantic model's Refresh history Correct answer.
    2. In the OneLake catalog
    3. In the Monitoring hub pipeline tab
    4. In the Capacity Metrics app

    Explanation

    Each semantic model exposes a Refresh history page that lists past refreshes with status, start and end times, and failure messages. This is the fastest way to confirm whether scheduled refreshes are succeeding and how long they take.

  137. Question 137Monitor and optimize an analytics solution

    Which Fabric item automatically takes an action, such as sending an email, when a condition in streaming data is met?

    1. A Reflex (Data Activator) item Correct answer.
    2. A pipeline with a scheduled trigger
    3. A KQL queryset
    4. An Eventstream with a derived stream

    Explanation

    Reflex, the Data Activator experience in Fabric, monitors data for defined conditions and fires actions such as emails or Teams messages when those conditions are met. Pipelines and Eventstreams move or transform data but do not evaluate business conditions and act on them.

  138. Question 138Monitor and optimize an analytics solution

    What does enabling autoscale for Spark in a Fabric workspace do?

    1. It adds or removes Spark nodes based on demand Correct answer.
    2. It moves Spark jobs to a different region
    3. It restarts failed Spark sessions automatically
    4. It raises the capacity SKU automatically when throttling occurs

    Explanation

    Spark autoscale dynamically adjusts the number of nodes allocated to a session between a minimum and maximum bound as the workload's demand changes. It affects node count, not the capacity SKU or region, and it does not retry failures.

  139. Question 139Monitor and optimize an analytics solution

    Sustained overuse of a Fabric capacity leads to which behavior for interactive requests?

    1. They are routed to a backup capacity
    2. They run at a higher priority
    3. They are delayed or rejected Correct answer.
    4. They are converted into background jobs

    Explanation

    When a capacity stays above its allowed utilization, Fabric applies throttling stages that first delay interactive requests and later reject them. Background operations are affected at a later stage. No automatic rerouting or priority escalation occurs.

  140. Question 140Monitor and optimize an analytics solution

    Which workspace feature shows how data moves between items such as lakehouses, notebooks, and reports?

    1. The workspace settings pane
    2. The lineage view Correct answer.
    3. The Real-Time hub
    4. The Monitoring hub

    Explanation

    The lineage view renders a graph of upstream and downstream dependencies across workspace items, which helps you assess the impact of a change or a failure. It answers 'what depends on this item' rather than 'did this item run successfully'.

  141. Question 141Monitor and optimize an analytics solution

    On a capacity that applies smoothing, how is the compute consumed by a short, intensive Spark job accounted for?

    1. It is charged in full at the time it ran
    2. It is smoothed across a 24-hour window Correct answer.
    3. It is billed only to the owning workspace
    4. It is excluded from capacity billing

    Explanation

    Smoothing spreads the compute consumed by an operation over a 24-hour window so that short spikes do not immediately push a capacity into overage. Because of smoothing, a burst of usage shows as a modest increase in utilization across the day rather than a single spike.

  142. Question 142Monitor and optimize an analytics solution

    What does bursting allow a Fabric capacity to do?

    1. To allow a capacity to permanently exceed the limits of its purchased SKU
    2. To queue background jobs until the capacity becomes idle again
    3. To replicate data to a secondary region during periods of peak demand
    4. To let a workload briefly exceed its share when capacity is idle Correct answer.

    Explanation

    Bursting lets a workload temporarily use more than its proportional share, up to a limit, while the capacity has spare resources, and the borrowed time is paid back later through smoothing. This absorbs short spikes without immediate throttling.

  143. Question 143Monitor and optimize an analytics solution

    Which Fabric Warehouse feature helps you find the queries that consume the most resources over a period?

    1. The Monitoring hub
    2. The Refresh history
    3. Query insights Correct answer.
    4. The lineage view

    Explanation

    Query insights in Fabric Warehouse surfaces aggregated query history, including duration, CPU, and other resource measures, so you can rank the most expensive queries and tune them. It is purpose-built for SQL workload analysis rather than generic run tracking.

  144. Question 144Monitor and optimize an analytics solution

    A Direct Lake semantic model starts returning slower results, and query plans show a fallback to SQL. What most likely happened?

    1. The capacity hosting the model was paused by an administrator
    2. The model hit a guardrail and fell back to DirectQuery mode Correct answer.
    3. The Delta tables backing the model were deleted from the lakehouse
    4. The storage mode of the model was changed from Direct Lake to Import

    Explanation

    Direct Lake models read Delta Parquet files directly, but when the data exceeds guardrails such as row or file counts, or when certain unsupported features are used, the model falls back to DirectQuery for affected queries. That fallback explains both the change in the plan and the slower response.

  145. Question 145Monitor and optimize an analytics solution

    What performance benefit does V-Order provide for Parquet files in OneLake?

    1. Faster reads for engines such as Power BI and Spark Correct answer.
    2. Compression of the transaction log
    3. Smaller capacity SKUs at the same workload
    4. Automatic schema evolution for Delta tables

    Explanation

    V-Order is a write-time layout optimization that improves how Parquet data is read by engines including Power BI and Spark, which reduces read time for Direct Lake and Spark queries. It changes the internal organization of the files, not the capacity size or table schema.

  146. Question 146Monitor and optimize an analytics solution

    A Delta table in a lakehouse consists of thousands of small files, and queries against it are slow. Which maintenance action should you run?

    1. OPTIMIZE to compact the files Correct answer.
    2. A full refresh of the semantic model
    3. ANALYZE TABLE to rebuild statistics
    4. VACUUM with a retention period of zero

    Explanation

    OPTIMIZE compacts many small Parquet files into fewer, larger files, which reduces per-file overhead during scans and improves read performance. It can also apply V-Order so downstream engines read the data more efficiently.

  147. Question 147Monitor and optimize an analytics solution

    A Spark job shows a very large shuffle during a join between a 2 GB table and a 5 MB table. Which change most improves performance?

    1. Broadcast the small table in the join Correct answer.
    2. Disable adaptive query execution
    3. Increase the number of executor cores
    4. Write the result as a single Parquet file

    Explanation

    When one side of a join is small, broadcasting it to every executor eliminates the shuffle of the large table and avoids the network cost of repartitioning both sides. This is the standard remedy when the broadcast threshold permits the small table to be shipped.

  148. Question 148Monitor and optimize an analytics solution

    Which feature helps you confirm that an Eventstream is receiving and delivering data as expected?

    1. Data insights and metrics on the Eventstream Correct answer.
    2. The lineage view of the workspace that contains the stream
    3. The Refresh history of the semantic model
    4. The Capacity Metrics app item page

    Explanation

    Eventstream provides data insights and metrics such as input and output event counts, errors, and backlog that show whether data is flowing through each node. These measures let you distinguish a source problem from a downstream processing problem.

  149. Question 149Monitor and optimize an analytics solution

    You need to investigate why records were not loaded into a KQL database. Which command shows the ingestion failures?

    1. .show tables
    2. .show database schema
    3. .show queries
    4. .show ingestion failures Correct answer.

    Explanation

    The .show ingestion failures control command returns recent ingestion operations that failed along with the failure reason, which is the direct way to diagnose rejected records. It complements ingestion metrics by giving per-operation detail.

  150. Question 150Monitor and optimize an analytics solution

    When a Fabric capacity remains above its limit for an extended period, which throttling stage occurs first?

    1. Background operations are rejected
    2. Interactive requests are rejected
    3. Interactive requests are delayed Correct answer.
    4. All operations are suspended for 24 hours

    Explanation

    Fabric throttles in progressive stages: interactive requests are first delayed, then rejected, and background operations are rejected at a later stage. The progression protects interactive responsiveness for as long as possible before harsher measures apply.

  151. Question 151Monitor and optimize an analytics solution

    You enable workspace monitoring to collect logs and metrics for the items in a workspace. Where is that collected data stored?

    1. In the workspace's default lakehouse as CSV files
    2. In the capacity's internal audit store
    3. In an Azure Log Analytics workspace
    4. In a KQL database in an eventhouse Correct answer.

    Explanation

    Workspace monitoring writes item logs and metrics into a KQL database hosted in an eventhouse, which you then query with KQL or visualize in Real-Time dashboards. This makes the telemetry available for joins with your own data and for long-term retention.

  152. Question 152Monitor and optimize an analytics solution

    A notebook run fails only when it processes the largest day of data. Which approach best diagnoses the failure?

    1. Review the capacity metrics app for compute overage during the run
    2. Inspect the Spark UI stage and task metrics for the failing run Correct answer.
    3. Check the lineage view of the lakehouse for upstream changes
    4. Rerun the notebook at a larger capacity SKU to see whether it succeeds

    Explanation

    The Spark UI shows which stage failed, per-task metrics such as spill and shuffle size, and executor logs, which is where the root cause of a data-volume-dependent failure is found. Capacity overage and size increases do not reveal the underlying fault.

  153. Question 153Monitor and optimize an analytics solution

    A semantic model refresh is slow and the model is very large. Which change most directly reduces refresh duration?

    1. Enable query caching on the capacity
    2. Remove unused columns and tables from the model Correct answer.
    3. Increase the SKU of the capacity
    4. Convert the model to DirectQuery storage mode

    Explanation

    Refresh time scales with the volume of data processed, so eliminating unused tables and columns reduces the data that must be read, compressed, and written. Capacity and caching changes affect query-time behavior rather than the amount of data processed during refresh.

  154. Question 154Monitor and optimize an analytics solution

    A pipeline that copies a large table takes far longer than expected. Which action most improves throughput?

    1. Disable fault tolerance in the copy activity
    2. Increase copy parallelism for the copy activity Correct answer.
    3. Schedule the pipeline to run outside business hours
    4. Reduce the data integration units of the copy activity

    Explanation

    Raising the degree of copy parallelism lets the copy activity move multiple partitions or file ranges concurrently, which shortens elapsed time for large volumes. Data integration units and parallelism are the primary tuning levers for copy throughput.

  155. Question 155Monitor and optimize an analytics solution

    Users report slow reports only during business hours on a capacity shared with many workspaces. What is the most likely cause?

    1. The semantic model refreshes run at night
    2. The workspace is assigned to a trial capacity
    3. The lakehouse files are compacted in the background
    4. Interactive workloads compete for capacity Correct answer.

    Explanation

    On a shared capacity, concurrent interactive workloads consume the same compute pool, so peak-hour contention degrades response times while off-peak usage is unaffected. Investigating per-item consumption in the Capacity Metrics app confirms which items drive the peak.

  156. Question 156Monitor and optimize an analytics solution

    You must identify which workspace consumed the most capacity over the past seven days. What should you do?

    1. Use the Capacity Metrics app with a 7-day window and the workspace table Correct answer.
    2. Open the Spark UI history server and review completed applications
    3. Open the lineage view of each workspace and count downstream items
    4. Filter the Monitoring hub to the last 7 days and group by item

    Explanation

    The Capacity Metrics app retains a rolling window of utilization data and presents it by workspace and by item, so selecting a 7-day range and reading the workspace-level table gives the comparison you need. The Monitoring hub and Spark UI report run outcomes rather than comparative consumption.

  157. Question 157Monitor and optimize an analytics solution

    A Spark join produces one task that runs far longer than all the others because of data skew. Which setting addresses this?

    1. spark.sql.autoBroadcastJoinThreshold set to a value of -1
    2. spark.sql.shuffle.partitions set to 1
    3. Adaptive query execution with skew join handling Correct answer.
    4. spark.executor.memory set to a smaller value

    Explanation

    Adaptive query execution can detect skewed partitions at runtime and split them into smaller sub-tasks so no single task dominates the stage. Setting the broadcast threshold to a negative value or forcing one shuffle partition makes the skew worse rather than better.

  158. Question 158Monitor and optimize an analytics solution

    A data engineer needs one place to review the run history, status, and duration of pipelines, notebooks, and dataflows across several workspaces. Which Fabric feature provides this?

    1. The Real-Time hub
    2. The Monitoring hub Correct answer.
    3. The OneLake catalog
    4. The workspace object browser

    Explanation

    The Monitoring hub aggregates run history, status, duration, and owner for items such as pipelines, notebooks, and dataflows across workspaces the user can access. It is the entry point for spotting a failed or unusually long run before drilling into item-specific detail.

  159. Question 159Monitor and optimize an analytics solution

    An administrator must analyze how much compute a Fabric capacity used over the past month and which items contributed to that usage. Which tool should the administrator use?

    1. The Monitoring hub
    2. The Spark UI
    3. The Capacity Metrics app Correct answer.
    4. A KQL queryset

    Explanation

    The Capacity Metrics app reports capacity unit consumption, throttling, and per-item and per-workspace breakdowns over time for a capacity. It is the standard tool for retrospective analysis of consumption trends.

  160. Question 160Monitor and optimize an analytics solution

    A notebook run is slow, and the engineer must inspect individual stage and task timing for that specific run. Which interface provides this detail?

    1. The Spark UI for the notebook run Correct answer.
    2. The capacity metrics app's Timepoint detail page
    3. The semantic model's refresh history page
    4. The pipeline run history

    Explanation

    Each notebook run executes as a Spark application, and the Spark UI exposes jobs, stages, tasks, shuffle volumes, and per-task durations. That level of detail is required to locate a slow stage or straggling task.

  161. Question 161Monitor and optimize an analytics solution

    A report shows stale data after an overnight load. Where can an engineer confirm whether the semantic model refresh completed and how long each step took?

    1. In the Spark UI
    2. In the capacity metrics app
    3. In the refresh history of the semantic model Correct answer.
    4. In a KQL queryset

    Explanation

    Refresh history records each refresh attempt with its status, start time, duration, and per-step detail, which is how a failed or partially completed refresh is confirmed.

  162. Question 162Monitor and optimize an analytics solution

    A report page takes several seconds to render. Which Power BI Desktop tool records the duration of each visual and its associated DAX query so the slowest element can be identified?

    1. Performance Analyzer Correct answer.
    2. The relationship diagram in Model view
    3. The filters pane of the report page
    4. The column statistics in Data view

    Explanation

    Performance Analyzer captures the time spent by each visual to render and to run its DAX query, letting the author isolate the slowest element on the page.

  163. Question 163Monitor and optimize an analytics solution

    A pipeline run took far longer than the previous run. Which view in the run history shows the start time and duration of each activity so the bottleneck can be spotted?

    1. The workspace item list
    2. The schedule trigger settings of the pipeline
    3. The JSON definition of the pipeline
    4. The Gantt view of the pipeline run Correct answer.

    Explanation

    The Gantt view plots each activity of a run on a timeline with its start, end, and duration, which makes a single long-running or blocking activity easy to identify.

  164. Question 164Monitor and optimize an analytics solution

    An engineer wants a notification when a pipeline's measured run duration exceeds 30 minutes. Which Fabric item can evaluate the condition and send the notification?

    1. An Eventstream with a derived stream
    2. A pipeline with a Wait activity
    3. A KQL queryset over the monitoring eventhouse
    4. A Data Activator rule on the metric Correct answer.

    Explanation

    Data Activator evaluates conditions on changing data or metrics and can trigger notifications or actions when a threshold is crossed. It is the Fabric item designed for alerting on monitored values.

  165. Question 165Monitor and optimize an analytics solution

    Compute consumption for Fabric workloads is reported in which unit?

    1. Request units per minute (RU/min)
    2. DTU seconds
    3. Capacity units (CU-s) Correct answer.
    4. vCore hours

    Explanation

    Fabric measures consumption in capacity units, reported per timepoint as CU seconds, and compares that consumption against the capacity's purchased limit.

  166. Question 166Monitor and optimize an analytics solution

    After a nightly load adds 50 million rows to a Fabric Warehouse table, several queries that previously ran quickly now scan far more data than expected. Which action most directly improves the query plans?

    1. Replace the warehouse tables with shortcuts
    2. Enable bursting on the capacity
    3. Move the queries to a KQL queryset
    4. Create statistics on the join and filter columns Correct answer.

    Explanation

    Fabric Warehouse does not create statistics automatically, so the optimizer can choose poor plans after large data changes. Creating statistics on the columns used in joins and predicates gives the optimizer the distribution information it needs.

  167. Question 167Monitor and optimize an analytics solution

    A Spark job joins two 100 MB tables and runs thousands of very small shuffle tasks, adding scheduling overhead. Which change reduces that overhead?

    1. Reduce spark.sql.shuffle.partitions to produce fewer, larger partitions Correct answer.
    2. Sort each input table before the join
    3. Increase spark.executor.cores so each executor runs more tasks concurrently
    4. Disable V-Order on the target table

    Explanation

    The shuffle partition count defaults to a value sized for much larger data, so small datasets produce many tiny tasks whose scheduling cost exceeds their work. Lowering the partition count (or letting adaptive execution coalesce partitions) produces fewer, better-sized tasks.

  168. Question 168Monitor and optimize an analytics solution

    Several small notebooks run every 10 minutes in the same workspace, and each starts a new Spark session, adding startup latency. Which setting reduces that overhead?

    1. Enable high concurrency mode for the notebooks Correct answer.
    2. Set spark.sql.adaptive.enabled to true
    3. Chain the notebooks in one pipeline with sequential dependencies
    4. Increase the capacity SKU size

    Explanation

    High concurrency mode lets multiple notebooks in the same workspace share a single Spark session, so a new session does not have to start for each notebook.

  169. Question 169Monitor and optimize an analytics solution

    A Direct Lake semantic model returns an error stating that a table has not been framed, and queries against that table fail. Which action resolves the issue?

    1. Refresh the semantic model so the table's Delta files are framed Correct answer.
    2. Enable V-Order on the lakehouse
    3. Switch the model to import mode for that table
    4. Re-create the OneLake shortcut that points to the Delta table

    Explanation

    Direct Lake serves queries by framing Delta table data into memory, and a newly added table or column is not queryable until framing occurs. Refreshing the semantic model performs that framing.

  170. Question 170Monitor and optimize an analytics solution

    A capacity exceeds its limit for several hours in a day and interactive operations are throttled. After usage falls below the limit, which behavior should be expected?

    1. All future background operations are blocked for 30 days
    2. The capacity returns to full throughput immediately with no lasting effect
    3. Overage is carried forward and capacity stays reduced until the debt is repaid Correct answer.
    4. The capacity is automatically upgraded to the next SKU and billed for the higher size

    Explanation

    Fabric uses smoothed consumption and records an overage deficit that is carried forward, so the capacity continues to operate with reduced available compute until the debt is paid off.

  171. Question 171Monitor and optimize an analytics solution

    A KQL query over a large Eventhouse table takes minutes, and the query only needs the last two hours of data. Which change most reduces the data scanned?

    1. Add a take operator at the end of the query
    2. Sort the table by timestamp before applying the other operators in the query
    3. Increase the retention period of the table
    4. Start the query with a filter on the timestamp column for the last two hours Correct answer.

    Explanation

    KQL pushes a timestamp filter down to storage so extents that fall outside the requested window are skipped entirely. Applying that filter first minimizes the data that later operators must process.

  172. Question 172Monitor and optimize an analytics solution

    Which KQL database policy keeps recently ingested records on local SSD storage so that queries over recent data run faster?

    1. The update policy
    2. The merge policy
    3. The retention policy
    4. The caching policy Correct answer.

    Explanation

    The caching policy defines the hot window during which data stays on local SSD, which makes recent-data queries faster. Retention controls how long data is kept, and the update policy runs transformations on ingest.

  173. Question 173Monitor and optimize an analytics solution

    A mirrored database must be checked for replication latency and for tables that failed to replicate. Which feature should the engineer use?

    1. The refresh history of the semantic model
    2. The mirroring status view for the database Correct answer.
    3. The query insights page of the SQL analytics endpoint
    4. The Spark UI of the mirroring workspace

    Explanation

    A mirrored database exposes a monitoring view that reports overall replication state, latency, and per-table status, including tables that are not replicating.

  174. Question 174Monitor and optimize an analytics solution

    A notebook fails with a driver out-of-memory error while processing a 40 GB table. The code calls collect() on a DataFrame and then loops over the rows in Python. Which change best fixes the root cause?

    1. Set spark.sql.shuffle.partitions to 1
    2. Enable high concurrency mode for the notebook
    3. Increase the number of executors so more partitions are processed in parallel
    4. Write the results to a lakehouse table with Spark instead of collecting them Correct answer.

    Explanation

    collect() moves every row of the result into driver memory, so the driver must hold the whole dataset. Keeping the processing distributed and writing the output with Spark avoids moving the data to a single node.

  175. Question 175Monitor and optimize an analytics solution

    Which Spark configuration lets Spark merge small shuffle partitions at runtime based on the actual data size?

    1. spark.sql.adaptive.enabled, which enables adaptive query execution Correct answer.
    2. spark.sql.files.maxPartitionBytes, which controls input file splits
    3. spark.sql.autoBroadcastJoinThreshold, which controls broadcast joins
    4. spark.executor.memory, which sets the memory per executor

    Explanation

    Adaptive query execution re-optimizes the physical plan at runtime using real statistics and coalesces contiguous small shuffle partitions into larger ones.

  176. Question 176Monitor and optimize an analytics solution

    An engineer must see which queries are currently executing in a Fabric Warehouse, including their elapsed time and status. Which object should be queried?

    1. The sys.objects catalog view
    2. An Eventstream of the workspace
    3. The sys.dm_exec_requests view Correct answer.
    4. The Delta log of the warehouse table

    Explanation

    Dynamic management views expose live session and request state, so sys.dm_exec_requests returns currently running requests with their status and elapsed time.

  177. Question 177Monitor and optimize an analytics solution

    After workspace monitoring is enabled, an engineer must analyze the collected logs with filtering by item and time. Which tool is designed for this?

    1. The capacity metrics app's Timepoint page
    2. The Spark UI's SQL tab
    3. A Power BI report over the warehouse's system views
    4. A KQL queryset over the monitoring eventhouse Correct answer.

    Explanation

    Workspace monitoring stores collected logs and metrics in an eventhouse, and a KQL queryset is the query tool for exploring that data with filters, aggregations, and time ranges.

  178. Question 178Monitor and optimize an analytics solution

    A dashboard runs the same aggregation over a billion-row Eventhouse table every few minutes, and each run scans the raw data. Which feature most reduces the query cost while keeping results current?

    1. A larger Eventhouse cluster with more compute cores
    2. A higher result-set size limit for the query
    3. A materialized view over the aggregation Correct answer.
    4. A retention policy set to 365 days

    Explanation

    A materialized view stores the precomputed aggregation and is incrementally updated as new data arrives, so repeated queries read the small view instead of rescanning raw data.

  179. Question 179Monitor and optimize an analytics solution

    The Spark UI shows a stage with very high shuffle read, long task durations, and a large share of task time spent in garbage collection. Which change most directly addresses the bottleneck?

    1. Increase executor memory to reduce garbage collection time Correct answer.
    2. Replace the join with a cross join
    3. Enable high concurrency mode for the notebooks
    4. Reduce the number of shuffle partitions to ten for the stage

    Explanation

    Garbage collection dominating task time after a large shuffle read indicates memory pressure, because tasks hold many objects that cannot be reclaimed quickly. Giving executors more memory, or reducing the data each task holds, lowers GC overhead.

  180. Question 180Monitor and optimize an analytics solution

    During an overage period, interactive operations were throttled while background operations continued to complete. Which comparison in the Capacity Metrics app best explains why interactive requests were affected first?

    1. The interactive versus background consumption at each timepoint Correct answer.
    2. The count of semantic model refreshes versus the count of pipeline runs
    3. The storage used by lakehouses versus the storage used by warehouses
    4. The number of workspaces assigned to this capacity versus other capacities

    Explanation

    Fabric smooths background and interactive consumption over different windows, so their shares at a given timepoint determine which limit is reached first. Comparing those shares shows why short interactive requests were throttled while smoothed background work proceeded.

  181. Question 181Monitor and optimize an analytics solution

    One task in a stage runs several times longer than its peers because a single executor is slow, while tasks on healthy executors finish quickly. Which Spark setting mitigates this?

    1. spark.speculation, which starts a duplicate attempt for slow tasks Correct answer.
    2. spark.sql.adaptive.enabled, which re-optimizes the plan at runtime
    3. spark.sql.files.maxPartitionBytes, which controls file read sizes
    4. spark.executor.instances, which sets how many executors are requested

    Explanation

    Speculative execution detects tasks running noticeably slower than the median and launches duplicate attempts on other executors, keeping the first result to finish. It targets stragglers caused by a slow node rather than by data distribution.

  182. Question 182Monitor and optimize an analytics solution

    An administrator must find which KQL queries against an Eventhouse consumed the most CPU during the last day. Which approach provides this?

    1. .show queries, which lists recent queries with resource statistics Correct answer.
    2. The capacity metrics item page for the eventhouse
    3. The ingestion failures view of the database
    4. The Spark UI SQL tab for the eventhouse

    Explanation

    The .show queries command returns recently executed queries together with their resource consumption statistics, so they can be sorted to find the heaviest consumers.

  183. Question 183Monitor and optimize an analytics solution

    An administrator must review the run history of pipelines, notebooks, and dataflows across multiple workspaces from a single location. Which Fabric feature provides this consolidated view?

    1. Monitoring hub Correct answer.
    2. Capacity Metrics app
    3. Deployment pipelines
    4. Workspace monitoring

    Explanation

    The Monitoring hub aggregates recent activity runs for items such as pipelines, notebooks, and dataflows across every workspace the user can access, showing status, start time, and duration. The Capacity Metrics app reports compute consumption at the capacity level rather than per-item run history.

  184. Question 184Monitor and optimize an analytics solution

    An administrator needs to identify which items consumed the most capacity compute yesterday and when usage exceeded the capacity limit. Which Fabric item should be opened?

    1. Real-Time hub
    2. OneLake catalog
    3. Monitoring hub
    4. Capacity Metrics app Correct answer.

    Explanation

    The Capacity Metrics app surfaces capacity-level telemetry such as utilization over time, throttling periods, and the top items by compute consumption. It is the designated tool for understanding how a capacity's resources were used.

  185. Question 185Monitor and optimize an analytics solution

    A Spark engineer must examine per-stage task durations, shuffle read and write sizes, and garbage collection time for a slow notebook run. Which Spark UI page provides these details?

    1. SQL tab
    2. Stages tab Correct answer.
    3. Executors tab
    4. Environment tab

    Explanation

    The Stages tab summarizes each stage's tasks and reports shuffle read and write metrics together with task-level timing, including the share of time spent in garbage collection. The Executors tab gives per-executor aggregate resource figures rather than stage-level task detail.

  186. Question 186Monitor and optimize an analytics solution

    Which policy on a KQL database determines how long records remain available for querying before they are automatically deleted?

    1. Caching policy
    2. Update policy
    3. Sharding policy
    4. Retention policy Correct answer.

    Explanation

    The retention policy defines the retention period, after which data is removed from the database and can no longer be queried. The caching policy governs how long data stays on local SSD for faster access and does not control deletion.

  187. Question 187Monitor and optimize an analytics solution

    A team must be notified automatically, with an email sent, when a KQL query returns a value above a defined threshold. Which Fabric item supports this?

    1. Notebook
    2. Eventstream
    3. Activator Correct answer.
    4. Data pipeline

    Explanation

    Activator evaluates conditions against data on a schedule or continuously and can trigger actions such as sending an email or a Teams message when a rule is met. The other items move or transform data but do not provide rule-based alerting on query results.

  188. Question 188Monitor and optimize an analytics solution

    Queries against a Fabric Warehouse table with a highly selective filter run slowly because the optimizer estimates far more rows than are actually returned. Which action most directly improves the plan?

    1. Wrap the query in a stored procedure
    2. Disable V-Order for the table
    3. Increase the capacity SKU
    4. Create or update statistics on the filtered columns Correct answer.

    Explanation

    The query optimizer uses column statistics to estimate cardinality, and poor estimates produce suboptimal join orders and scan strategies. Creating or updating statistics on the filtered columns gives the optimizer the distribution information it needs.

  189. Question 189Monitor and optimize an analytics solution

    A Spark job reads a large Parquet table stored as many files and creates thousands of tiny tasks, and the engineer wants to control how much data each task reads. Which configuration should be adjusted?

    1. spark.sql.autoBroadcastJoinThreshold
    2. spark.sql.shuffle.partitions
    3. spark.executor.cores
    4. spark.sql.files.maxPartitionBytes Correct answer.

    Explanation

    spark.sql.files.maxPartitionBytes sets the maximum number of bytes packed into a single input partition when reading files, so raising it produces fewer and larger read tasks. It applies to file scan partitioning rather than to shuffle or join behavior.

  190. Question 190Monitor and optimize an analytics solution

    What is the primary benefit of writing Delta tables with V-Order enabled in Fabric Spark?

    1. It encrypts table data at rest
    2. It merges small files during every write
    3. It speeds Parquet reads through sorting and compression Correct answer.
    4. It removes the need for Delta checkpoints

    Explanation

    V-Order applies an optimized sort and compression layout when Parquet files are written, which lets Parquet-based readers, including Fabric query engines, read the data faster. It does not replace compaction, checkpointing, or encryption.

  191. Question 191Monitor and optimize an analytics solution

    Fabric applies 24-hour smoothing to background operations. What does this mean for a large notebook run that consumes a burst of capacity?

    1. Its usage is spread across the day rather than charged at once Correct answer.
    2. The operation is retried later on a smaller cluster
    3. The cost is billed to the workspace owner instead
    4. Its usage is excluded from utilization calculations

    Explanation

    Smoothing distributes the compute consumed by background operations over a 24-hour window so that a single spike does not instantly push the capacity into a throttled state. The consumption still counts toward utilization; it is only averaged over time.

  192. Question 192Monitor and optimize an analytics solution

    A Dataflow Gen2 query pulls a large table from SQL Server, applies transformations, and the transformations run row by row after extraction. Which change improves refresh performance?

    1. Enable V-Order on the destination lakehouse
    2. Land the raw table in a notebook first
    3. Increase the number of gateway nodes
    4. Keep transformations foldable so the source filters and aggregates Correct answer.

    Explanation

    When transformations remain foldable, they are translated into native queries executed by the source, so far fewer rows are extracted. Steps that break folding force the engine to materialize and process rows locally.

  193. Question 193Monitor and optimize an analytics solution

    An engineer must analyze queries that finished in a Fabric Warehouse over the past week, including duration and CPU time for each request. Which object should be queried?

    1. sys.dm_exec_sessions
    2. sys.dm_exec_connections
    3. queryinsights.exec_requests_history Correct answer.
    4. sys.partitions

    Explanation

    queryinsights.exec_requests_history retains completed request records with timing, status, and resource metrics, which makes it suitable for historical analysis. The sys.dm_exec_* views describe the current session and connection state rather than a week of completed queries.

  194. Question 194Monitor and optimize an analytics solution

    A Spark join stage has one task that processes far more rows than the others because of a skewed join key. Which change addresses the skew?

    1. Enable adaptive query execution skew join handling Correct answer.
    2. Increase spark.sql.shuffle.partitions sharply
    3. Broadcast the larger table to every executor
    4. Cache both joined tables before the join

    Explanation

    Adaptive query execution can detect skewed shuffle partitions and split the oversized partition into smaller sub-tasks so that work is distributed evenly. Simply raising the partition count leaves the skewed key in a single partition.

  195. Question 195Monitor and optimize an analytics solution

    A PySpark notebook applies a row-at-a-time Python UDF to 500 million rows, and the job runs slowly. Which change improves throughput most?

    1. Increase the driver memory
    2. Cache the DataFrame before applying the UDF
    3. Write intermediate results to Parquet between calls
    4. Replace the Python UDF with built-in functions or a vectorized pandas UDF Correct answer.

    Explanation

    A standard Python UDF forces row-by-row serialization between the JVM and the Python worker, which dominates runtime at large scale. Built-in Spark SQL functions or vectorized pandas UDFs process data in batches and avoid most of that overhead.

  196. Question 196Monitor and optimize an analytics solution

    A lakehouse table is written by many small streaming appends and now contains thousands of small files. Which command compacts them into larger files?

    1. DESCRIBE HISTORY
    2. CACHE TABLE
    3. OPTIMIZE Correct answer.
    4. VACUUM

    Explanation

    OPTIMIZE performs bin compaction on a Delta table, merging small files into larger ones so that readers open fewer files and scan more efficiently. VACUUM instead removes unreferenced old files after the retention period.

  197. Question 197Monitor and optimize an analytics solution

    A Dataflow Gen2 query pulls from an on-premises SQL Server, applies transformations that the source cannot fold, and is referenced by three other queries in the same dataflow. Which configuration improves refresh performance?

    1. Change the destination to a warehouse
    2. Disable the on-premises data gateway
    3. Enable staging to persist intermediate results Correct answer.
    4. Raise the dataflow's request timeout

    Explanation

    Staging writes intermediate query results to a lakehouse so that non-foldable transformations are computed once instead of being re-evaluated for each downstream query. This is most beneficial when several queries reuse the same expensive source.

  198. Question 198Monitor and optimize an analytics solution

    Queries on a large Delta table filter on a customer column, but the table is physically ordered by date and file skipping does not eliminate files. Which operation improves skipping for those filters?

    1. Run VACUUM with a zero-hour retention
    2. Cache the table with CACHE TABLE
    3. Run OPTIMIZE with Z-ORDER on the filter column Correct answer.
    4. Disable V-Order for the table

    Explanation

    Z-ORDER clusters rows with similar values of the specified columns into the same files, so the per-file minimum and maximum statistics become tighter and more files can be skipped. Physical ordering by date leaves the customer values scattered across every file.

  199. Question 199Monitor and optimize an analytics solution

    A Copy activity moves 4 TB from an on-premises SQL Server to a lakehouse and runs for many hours using a single stream. Which change increases throughput?

    1. Increase copy parallelism and DIU count Correct answer.
    2. Reduce the copy batch size to 1,000 rows
    3. Set fault tolerance to skip incompatible rows
    4. Enable staging in a dataflow

    Explanation

    Copy activity throughput scales with the number of parallel copy connections and the data integration units allocated to the activity, allowing the source and sink to be read and written concurrently. A single stream underuses the available bandwidth and source capability.

  200. Question 200Monitor and optimize an analytics solution

    A Spark shuffle stage runs 200 tasks, each handling about 2 GB, and tasks spill heavily to disk. Which adjustment is most appropriate?

    1. Reduce spark.sql.shuffle.partitions to 50
    2. Raise spark.sql.autoBroadcastJoinThreshold to 1 GB
    3. Disable adaptive query execution
    4. Increase spark.sql.shuffle.partitions so each task handles roughly 100-200 MB Correct answer.

    Explanation

    Shuffle partitions that are far larger than the recommended target size force each task to hold large intermediate results in memory and spill to disk. Raising the partition count produces smaller partitions that fit in memory and reduce spilling.

Fabric DP-700 flashcards

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

OneLake
The single logical data lake for a tenant, built on ADLS Gen2 with a hierarchical namespace. Every Fabric item stores its data in OneLake under workspace and item folders.
Fabric workspace roles
Admin, Member, Contributor, and Viewer. Contributors create and edit items, Viewers are read-only, and only Admins can manage workspace permissions and membership.
OneLake shortcut
An object that points to data stored elsewhere, such as ADLS Gen2, Amazon S3, Dataverse, or another OneLake path, without copying it. Reads are served from the source.
Database mirroring
Continuous near real-time replication of an external source database into OneLake as Delta tables. Mirrored data is read-only and queryable by Spark, SQL, and Power BI.
Deployment pipeline
A Fabric tool that assigns workspaces to stages such as Development, Test, and Production and copies item definitions between them, with rules that override connections and parameters per stage.
SQL analytics endpoint
A read-only T-SQL surface created automatically with each lakehouse over its Delta tables. It supports SELECT and metadata queries but not data modification.
Fabric Warehouse item
A relational item supporting full T-SQL DML, multi-table transactions, stored procedures, and views, with data persisted in Delta format in OneLake.
Fabric capacity SKU
Capacity is purchased in capacity units with SKUs from F2 to F2048. A workspace must be assigned to a capacity before its items can execute.
Promotion versus certification
Promotion endorses an item as trustworthy and requires write permission on it. Certification requires approval from a tenant-designated group and marks organization-trusted data.
Fabric domains
A governance construct that groups workspaces into business areas, letting administrators delegate workspace assignment and control how data is discovered by users.
Sensitivity label inheritance
Microsoft Purview labels applied to a Fabric item are inherited by downstream items that Fabric can trace through lineage, so protection follows the data.
Copy job
A Fabric item for no-code bulk and incremental data movement between supported sources and destinations, supporting column mapping and change-based copying.
Dataflow Gen2
A Power Query Online-based item that extracts, transforms, and loads data. Output destinations include lakehouse tables, warehouse tables, and KQL databases.
Eventstream
A Fabric item that ingests streaming data from Event Hubs, IoT Hub, Kafka, or sample sources and routes it to lakehouse, KQL database, derived streams, or a custom endpoint.
Medallion architecture
A layered data design where bronze holds raw ingested data, silver holds cleansed and conformed data, and gold holds business-level aggregates used for reporting.
Delta transaction log
An ordered JSON log stored in the _delta_log folder that records every change to a Delta table, enabling ACID transactions, schema enforcement, and time travel.
OPTIMIZE command
A Delta operation that compacts many small files into fewer larger ones and can optionally apply Z-ORDER, improving read performance after frequent incremental writes.
VACUUM default retention
VACUUM deletes data files no longer referenced by the Delta log. The default retention threshold is seven days unless the retention check is explicitly disabled.
V-Order
A write-time Parquet optimization in Fabric that sorts and compresses row groups to accelerate reads, especially Power BI Direct Lake queries. It is enabled by default.
Watermark incremental load
A pattern that stores the maximum value of a column such as last-modified timestamp, then copies only rows newer than that watermark on each subsequent run.
ForEach activity
A pipeline activity that iterates over an array of items. Its batch count setting controls how many iterations execute in parallel during the run.
Queued versus streaming ingestion
Streaming ingestion writes KQL events with low latency but limited throughput; queued ingestion batches data for higher throughput with seconds-to-minutes delay.
Monitoring hub
A Fabric portal page listing activity runs, item runs, and refreshes across workspaces with status, duration, and error details, filterable by workspace, item, and time.
Capacity metrics app
A Power BI app that reports compute and storage consumption per workspace and item for a Fabric capacity, along with overload, throttling, and autoscale signals.
Capacity smoothing
A mechanism that spreads short compute spikes over a 24-hour window so brief peaks above the capacity limit do not immediately cause throttling.
Capacity bursting
A behavior that lets a workload temporarily consume more than its provisioned capacity by borrowing against future capacity, with the debt recovered later through smoothing.
Query insights
A Fabric Warehouse feature that surfaces the longest-running and most frequent queries with their execution plans and resource usage so they can be tuned.
Spark Advisor
Automated guidance surfaced after Fabric notebook runs that flags issues such as partition skew, oversized shuffles, or inefficient joins with suggested remediations.
Direct Lake fallback
When a semantic model cannot read Delta tables directly, for example due to unsupported data types or excessive files, it falls back to DirectQuery, which performs more slowly.
Result set caching
A warehouse feature that stores query results so identical repeated queries return from cache. It can be enabled or disabled per session with SET RESULTSET_CACHE.
Statistics in warehouse
Metadata describing column value distribution that the query optimizer uses to estimate row counts. Missing or stale statistics produce poor execution plans.
KQL materialized view
A KQL database object that stores the precomputed result of a summarization query over a source table and is kept current automatically as new data arrives.
Fabric throttling stages
Escalating responses to capacity overload: interactive delay first, then interactive rejection, then background rejection, applied once smoothed utilization exceeds capacity limits.

Fabric DP-700 study guide

Implement and manage an analytics solution

What this domain tests

You must choose the right Fabric item, wire it into a workspace and capacity, and lock it down. Know the difference between a lakehouse (Spark + Delta files, with a read-only SQL analytics endpoint), a warehouse (full T-SQL DDL/DML, no Spark), an eventhouse/KQL database (streaming, time-series), and a semantic model on top of any of them. OneLake is the single logical lake; shortcuts virtualize data (internal, ADLS Gen2, S3/GCS, Dataverse) without copying it.

Security, in layers

  • Workspace roles: Admin, Member, Contributor, Viewer. Contributors can create and publish items but cannot manage workspace permissions.
  • Item permissions and sharing: granting access to a single lakehouse or warehouse.
  • Granular data security (T-SQL): row-level security (RLS) via CREATE SECURITY POLICY, column-level security with GRANT/DENY, and dynamic data masking on warehouse columns.
  • Governance: sensitivity labels (Purview), endorsements (Promoted vs Certified), domains for data-mesh organization, and lineage view.

Lifecycle

Deployment pipelines (dev/test/prod with deployment rules and variable libraries) and Git integration move items between stages.

Traps

  1. Assuming RLS defined on a lakehouse's SQL analytics endpoint protects Spark readers - it does not; Spark reads the Delta files directly. Enforce RLS in the semantic model as well if needed.
  2. Confusing workspace role with item permission: a Viewer of the workspace may still be blocked from specific items.
  3. Thinking a lakehouse SQL endpoint allows UPDATE/DELETE - it is read-only.
  4. Forgetting that mirroring produces a read-only replica.

Rule of thumb

"Need T-SQL writes (INSERT/UPDATE/DELETE, stored procedures)? Warehouse. Need Spark, notebooks, and file-level control? Lakehouse. Need sub-second streaming queries? Eventhouse."

Ingest and transform data

The ingestion toolbox

Match the tool to the shape of the problem:

  • Copy activity / Copy job in a pipeline - bulk movement, best for large one-shot or scheduled loads; supports staging for warehouse loads.
  • Dataflow Gen2 - Power Query Online transformations with a destination (lakehouse, warehouse, KQL DB, Azure SQL); low-code, good for moderate volumes.
  • Shortcuts - zero-copy access; nothing lands in your storage.
  • Mirroring - continuous, near-real-time replication of databases (SQL Server, Azure SQL, Cosmos DB, Snowflake) into OneLake as Delta.
  • Eventstream - streaming ingestion into an eventhouse or lakehouse.
  • Notebooks (PySpark) - complex, code-first transformations; MERGE INTO for upserts into Delta.

Medallion pattern

Bronze (raw append), Silver (cleansed, deduplicated, conformed), Gold (business-level aggregates serving semantic models). Delta Lake gives ACID transactions, time travel, schema enforcement/evolution, and MERGE.

Pipelines vs dataflows

Pipelines orchestrate (activities, parameters, triggers: schedule, tumbling window, event). Dataflows do the transformation logic. A pipeline can invoke a dataflow and chain notebooks.

Traps

  1. Believing a shortcut copies data - it does not, and external shortcuts still need credentials/permissions at query time.
  2. Expecting to write to a mirrored database - it is read-only.
  3. Using Dataflow Gen2 for multi-terabyte transforms when Spark is the better fit.
  4. Forgetting that Delta MERGE needs a stable key, and that schema evolution must be enabled for new columns.

Rule of thumb

"If you can avoid moving the data, shortcut or mirror it. If a copy must exist as Delta, use Copy activity. If the logic is complex or large-scale, use a notebook."

Monitor and optimize an analytics solution

Where to look

  • Monitoring hub - one view of pipeline runs, notebook and Spark job activity, dataflow refreshes, and their status/duration.
  • Spark UI and Spark Advisor - stage/task skew, shuffle, executor utilization; Advisor surfaces concrete recommendations (partitioning, caching, broadcast joins).
  • Capacity Metrics app - capacity units (CU) consumption by item, and throttling stages: interactive delay, then interactive rejection, then background rejection. Watch the 24-hour smoothing effect before concluding a spike is a capacity problem.
  • Query insights / DMVs (warehouse) - slow T-SQL, repeated queries, missing statistics.

Storage and table tuning

Fabric writes Parquet with V-Order enabled by default, which speeds Direct Lake and Spark reads. Avoid small files: compact with OPTIMIZE (bin compaction, optional Z-ORDER on filter columns) and reclaim space with VACUUM (default 7-day retention). Partition only on high-cardinality, commonly filtered columns - over-partitioning creates the small-file problem you were trying to avoid. Target file sizes around 1 GB.

Compute tuning

Enable autoscaling and dynamic allocation for Spark, use high concurrency mode to share a session across notebooks, and set optimize write for partitioned sinks. Size capacity to the workload, then reduce waste: cancel idle sessions, schedule heavy jobs outside peak, and avoid full refreshes when incremental refresh suffices.

Direct Lake

Direct Lake reads Delta/Parquet directly from OneLake. If a query uses something it cannot translate (e.g., a SQL view, certain expressions) or exceeds guardrails, the model can fall back to DirectQuery - a classic exam trap. Check for views, unsupported data types, or overly large tables.

Rule of thumb

"Slow reads → check file sizes and run OPTIMIZE. Slow Direct Lake → check for fallback to DirectQuery. Slow everything → check the Capacity Metrics app for throttling."

Fabric DP-700 cheat sheet

Quick reference

Implement and manage an analytics solution

  • Workspace role order is Admin > Member > Contributor > Viewer. Contributor can create items and write data but cannot manage workspace permissions or add members.

  • OneLake uses a hierarchical path per item: onelake:////... ; every Fabric item stores its data under its workspace and item folder.

  • A workspace must be assigned to a Fabric capacity before its items run. SKUs range from F2 (smallest) to F2048, measured in capacity units.

  • Lakehouse = Spark + Delta tables + an auto-generated, read-only T-SQL SQL analytics endpoint. Warehouse = full T-SQL DML, multi-statement transactions, stored procedures, views.

  • Shortcuts store no data copy; deleting a shortcut never deletes the source. Shortcuts may chain but circular references are blocked.

  • Deployment pipelines map stages (Development/Test/Production) to workspaces and copy item definitions forward; deployment rules override per-stage connections and parameters. Not every item type is supported.

  • Promotion endorsement requires write permission on the item; certification requires approval by a tenant-designated security group. Both render as badges in search results.

  • Purview sensitivity labels applied to Fabric items are inherited by downstream items that Fabric can trace through lineage.

  • Git integration connects a workspace to an Azure Repos or GitHub branch; only supported item types are tracked, and changes flow via commit and update operations.

Quick reference

Ingest and transform data

  • Copy job is a standalone no-code item for bulk and incremental movement between sources and destinations; a Copy activity is the pipeline-level equivalent for orchestrated runs.

  • Dataflow Gen2 is built on Power Query Online and can write results to lakehouse tables, warehouse tables, KQL databases, and Azure SQL.

  • Core pipeline activities: Copy, Notebook, Dataflow, Stored Procedure, Lookup, Get Metadata, ForEach, If Condition, Until, Wait, Set Variable, Web.

  • Eventstream sources include Event Hubs, IoT Hub, Kafka, CDC feeds, and sample data; destinations include lakehouse, KQL database, derived stream, and custom endpoint.

  • Delta maintenance: OPTIMIZE compacts small files (optionally with Z-ORDER); VACUUM removes unreferenced files with a default retention threshold of seven days.

  • Incremental load patterns: watermark column comparison, Fabric CDC, or database mirroring - full reloads are reserved for initial or corrective loads.

  • Spark notebooks default to PySpark and can mix languages with cell magics such as %%pyspark, %%sql, and %%scala; attach a lakehouse to read and write Delta tables.

  • V-Order is a write-time Parquet optimization enabled by default in Fabric Spark; it improves read throughput, particularly for Power BI Direct Lake models.

  • Schema drift is handled in Delta by merge-schema options during writes; without it, new columns fail the write or are dropped.

Quick reference

Monitor and optimize an analytics solution

  • The Monitoring hub lists activity runs, item runs, and refreshes across workspaces with status, duration, and error text, filterable by workspace, item, and time range.

  • The Fabric Capacity Metrics app reports compute and storage consumption per workspace and item, plus overload and throttling signals for the capacity.

  • Capacity smoothing spreads short compute spikes across a 24-hour window so brief peaks above the limit do not immediately cause rejection.

  • Throttling escalates in stages: interactive delay, then interactive rejection, then background rejection, once smoothed utilization stays over capacity.

  • Warehouse tuning: create or update statistics on join and filter columns after large loads; the optimizer relies on them to estimate cardinality.

  • Query insights surfaces the longest-running and most frequent warehouse queries with plans and resource usage, and flags missing statistics.

  • Frequent small writes cause Direct Lake framing to degrade; run OPTIMIZE and keep V-Order enabled. When guardrails are exceeded, the model falls back to DirectQuery.

  • Spark tuning: enable autoscale and dynamic allocation, watch shuffle size and partition skew in the Spark UI, and read Spark Advisor hints after notebook runs.

  • KQL tuning: apply time filters first, use materialized views for repeated aggregations, and rely on query caching for repeated identical queries.

Fabric DP-700 key terms

OneLake

A single, tenant-wide logical data lake that holds all Fabric data as Delta Parquet. Each workspace gets its own folder inside it, and items reference that storage rather than keeping private copies.

Related: Shortcut, Lakehouse, Delta Parquet
Lakehouse

A Fabric item that pairs file storage with managed Delta tables and an auto-generated SQL endpoint, so the same data can be queried with Spark or T-SQL without duplication.

Related: SQL analytics endpoint, Delta Parquet, Lakehouse schema
Warehouse

A Fabric item offering a full T-SQL relational engine with multi-table transactions, views and stored procedures, storing its tables as Delta Parquet in OneLake.

Related: SQL analytics endpoint, Stored procedure, Row-level security
Workspace

A container that holds Fabric items such as lakehouses, notebooks, pipelines and reports, and is where you assign roles, attach capacity and connect source control.

Related: Fabric capacity, Deployment pipeline, Git integration
Fabric capacity

The pool of compute, measured in capacity units, that powers every workload in a workspace. It can be paused, resized, and throttles work when demand exceeds its size.

Related: Capacity metrics app, Smoothing, Workspace
Shortcut

An object in OneLake that points to data stored elsewhere, such as another lakehouse, ADLS Gen2, S3 or Dataverse, making it appear locally without being copied.

Related: OneLake, Mirroring, Lakehouse
Delta Parquet

An open table format made of Parquet data files plus a JSON transaction log that records commits, giving ACID guarantees and time travel. Fabric stores all OneLake data this way.

Related: OneLake, V-Order, VACUUM
SQL analytics endpoint

The read-only T-SQL surface automatically created for a lakehouse or mirrored database, exposing Delta tables as queryable tables and views without duplicating storage.

Related: Lakehouse, Warehouse, Mirroring
Direct Lake

A Power BI storage mode that reads Delta Parquet files straight from OneLake, avoiding an import copy while keeping import-like speed; it can fall back to DirectQuery when needed.

Related: Semantic model, V-Order, Lakehouse
Semantic model

The layer of tables, relationships, measures and security rules that reports query. It sits on top of a lakehouse, warehouse or other source and shapes how business users see data.

Related: Direct Lake, Row-level security
Lakehouse schema

A named grouping of tables inside a lakehouse that organizes objects into folders and lets permissions be managed at the schema level rather than table by table.

Related: Lakehouse, OneLake security roles
OneLake security roles

Role definitions scoped to folders, schemas or tables in OneLake that grant or deny read access for users and groups, applying consistently across Spark, SQL and Power BI.

Related: OneLake, Row-level security, Lakehouse schema
Row-level security

A rule that filters the rows a person can see based on their identity, implemented with a security predicate in T-SQL or with model roles in a semantic model.

Related: OneLake security roles, Warehouse
Deployment pipeline

A Fabric feature that promotes items through development, test and production workspaces, with rules that swap data source and parameter values at each stage.

Related: Workspace, Git integration
Git integration

Connecting a workspace to a Git repository so item definitions are versioned, changed on branches and synced between source control and the Fabric service.

Related: Workspace, Deployment pipeline
Medallion architecture

A layered data design where raw ingested data lands in bronze, is cleaned and conformed into silver, and is modeled into gold tables ready for reporting.

Related: Lakehouse, Data pipeline, Dataflow Gen2
Data pipeline

A Fabric item that orchestrates activities such as copy, notebook, stored procedure and conditional control flow, started by a schedule, event or manual run.

Related: Copy job, Notebook, Stored procedure
Copy job

A Fabric item focused on bulk or incremental data movement between a source and destination, requiring less configuration than a full orchestration pipeline.

Related: Data pipeline, Incremental load
Dataflow Gen2

A Power Query based item with a low-code graphical editor that extracts, transforms and loads data, writing results into lakehouses, warehouses or other supported destinations.

Related: Data pipeline, Medallion architecture
Notebook

An interactive document of Spark code cells in PySpark, Spark SQL, Scala or R, attached to a lakehouse, used for exploration, transformation and ad hoc analysis.

Related: Spark job definition, Spark UI, Lakehouse
Spark job definition

A Fabric item that runs a script or compiled artifact against a Spark cluster on a schedule, without the interactive, cell-by-cell experience of a notebook.

Related: Notebook, Data pipeline
Eventstream

A Fabric item that ingests streaming events from sources such as Event Hubs or Kafka, optionally transforms them in flight, and routes them to a destination.

Related: Eventhouse, Real-Time hub
Eventhouse

A real-time analytics store holding one or more KQL databases, optimized for high-volume streaming data queried with Kusto Query Language and shown in real-time dashboards.

Related: Eventstream, Real-Time hub
Real-Time hub

A central catalog in Fabric for discovering streaming sources, previewing data and subscribing to events so they can be consumed by eventstreams and other items.

Related: Eventstream, Eventhouse
Mirroring

Continuous replication of an external database such as SQL Server, Cosmos DB or Snowflake into OneLake, producing a near-real-time read-only copy without building pipelines.

Related: Shortcut, SQL analytics endpoint
Stored procedure

A saved T-SQL routine in a Fabric warehouse that accepts parameters and executes one or more statements, often invoked by a pipeline activity to load or transform data.

Related: Warehouse, Data pipeline
Incremental load

A pattern that loads only rows newer than a stored high-water mark instead of reloading a whole table, cutting runtime and compute cost on repeated runs.

Related: Copy job, Data pipeline
Monitor hub

The Fabric page listing activity across items, including pipeline runs, notebook runs and refreshes, with status, duration and drill-through details for troubleshooting failures.

Related: Capacity metrics app, Spark UI
Capacity metrics app

A Power BI app that reports capacity unit consumption, throttling and overload broken down by item and operation, used to right-size and tune a Fabric capacity.

Related: Fabric capacity, Smoothing
Smoothing

A capacity behavior that spreads the cost of background operations over time so short bursts are accounted for gradually rather than causing immediate throttling.

Related: Fabric capacity, Capacity metrics app
Spark UI

The diagnostic interface for a Spark session showing jobs, stages, tasks, shuffles and executor memory, used to spot skew, spills and slow or failing notebook runs.

Related: Notebook, Monitor hub
Query insights

A Fabric warehouse feature that surfaces the most frequent and most expensive queries along with their plans and resource usage so they can be tuned.

Related: Statistics, Warehouse
Statistics

Metadata describing how column values are distributed, used by the SQL engine to pick join orders and access paths; stale statistics often cause slow queries.

Related: Query insights, Warehouse
V-Order

A write-time layout optimization that sorts and compresses Parquet files so Fabric engines read them faster, particularly benefiting Direct Lake and SQL queries.

Related: Delta Parquet, Direct Lake, OPTIMIZE
OPTIMIZE

A Delta maintenance command that rewrites many small files into fewer larger ones, optionally sorting or clustering rows, improving read performance for downstream queries.

Related: V-Order, Z-order, VACUUM
VACUUM

A Delta command that deletes data files no longer referenced by the transaction log and older than a retention threshold, reclaiming storage and speeding file listing.

Related: Delta Parquet, OPTIMIZE
Z-order

A Delta layout technique that co-locates related values across several columns into the same files, letting filtered queries skip more data during reads.

Related: OPTIMIZE, Partitioning
Partitioning

Physically splitting a table into folders by a column such as date so that queries filtering on that column read only the relevant files instead of scanning everything.

Related: Z-order, OPTIMIZE

Fabric DP-700 study plan

8 weeks · About 8 hours per week

Week 1

Ingest and transform data

  • Provision a Fabric workspace and build a lakehouse, then load 2 distinct source types (a CSV file and an Azure SQL table) into the bronze layer

  • Create a Data pipeline with a Copy activity plus a parameterized ForEach loop, and configure a scheduled trigger

  • Build a Dataflow Gen2 that ingests and lightly cleanses a semi-structured source, then lands output in the lakehouse

  • Explain when to choose pipeline vs Dataflow Gen2 vs notebook vs Copy job for a given ingestion scenario, and write the decision down

Your checkpoint

A working workspace with 3 successful end-to-end pipeline runs recorded in run history, plus a one-page ingestion-pattern decision table

Week 2

Ingest and transform data & Implement and manage an analytics solution

  • Transform bronze data into silver using a Spark notebook with PySpark DataFrame operations and Delta writes

  • Implement a slowly changing dimension type 2 pattern (or upsert/merge) and verify row history is preserved

  • Model gold-layer tables with a star schema and validate joins against expected row counts

  • Create OneLake shortcuts to an external ADLS container and confirm the data is queryable without copying

  • Configure a shortcut or mirroring path for a second source and document the trade-offs

Your checkpoint

A medallion lakehouse (bronze/silver/gold) with a written data-quality check report showing reconciliation row counts for each layer

Week 3

Implement and manage an analytics solution

  • Design and implement workspace roles, item permissions, and row-level/column-level security on a gold table

  • Set up a deployment pipeline or Git integration and promote content from dev to test with parameterized connections

  • Configure sensitivity labels and endorse a certified item; document the governance choices

  • Query lakehouse data through the SQL analytics endpoint and compare results with a Spark query on the same table

Your checkpoint

A promoted workspace pair (dev to test) with a security matrix document listing each role, item, and granted access, verified by logging in as a restricted test user

Week 4

Implement and manage an analytics solution & Monitor and optimize an analytics solution

  • Wire a Real-Time/KQL database or Eventstream end-to-end and query it with KQL to confirm live ingestion

  • Add orchestration dependencies, retry logic, and failure-path handling to an existing pipeline

  • Enable basic monitoring: review the Monitoring hub and capture a failed-run investigation end-to-end

  • Document the end-to-end solution architecture with data flow, compute choices, and security boundaries

Your checkpoint

An architecture diagram plus a runbook covering the deployed solution, and a resolved (or deliberately simulated and diagnosed) pipeline failure with root cause written up

Week 5

Monitor and optimize an analytics solution

  • Interpret the Fabric Capacity Metrics app to identify the top throttling/smoothing consumers over a 7-day window

  • Diagnose a slow Spark notebook using the Spark UI (skew, shuffle, spill) and apply one fix

  • Diagnose a slow warehouse/SQL query using the query insights view and rewrite it for improvement

  • Configure an alert or Data Activator rule that fires on a pipeline failure or threshold breach, and prove it fires

Your checkpoint

A performance baseline report with before/after timing for 2 workloads and a screenshot-free written record of one alert that actually fired

Week 6

Monitor and optimize an analytics solution

  • Apply Delta optimization techniques (file compaction, partitioning strategy, V-Order/target file sizing) and measure the effect on query duration

  • Right-size compute: compare capacity SKU behavior, choose between scaling up vs pausing/resuming, and justify the cost trade-off

  • Automate monitoring by scheduling a notebook or pipeline that logs run durations and failures to a table

  • Build a small dashboard or Power BI report over the monitoring log and add 2 alert thresholds

Your checkpoint

A tuning log showing measurable improvement on at least 3 queries plus a reusable monitoring notebook committed to the workspace

Week 7

Ingest and transform data & Implement and manage an analytics solution & Monitor and optimize an analytics solution

  • Rebuild one weak-area component from scratch without notes to prove retention

  • Complete 40+ practice items across all three domains and log every miss with the reason (knowledge gap vs misread vs unknown feature)

  • Re-study only the logged gaps and re-test yourself on those items after 48 hours

  • Re-run the full solution end-to-end after making one schema change and confirm nothing breaks

Your checkpoint

An error log with 100% of missed items re-attempted, and a documented passing re-test rate of at least 85% on the previously missed set

Week 8

Implement and manage an analytics solution & Ingest and transform data & Monitor and optimize an analytics solution

  • Take 2 full-length timed practice assessments and score them against the 70% target per domain

  • Review every remaining flagged topic and convert each into a 5-bullet summary card

  • Walk through the official exam skill outline item-by-item and mark each as confident or shaky

  • Do a final dry run of the end-to-end solution: ingest, transform, secure, monitor, and tune in one sitting

Your checkpoint

Two timed practice scores at or above target, a completed skills-outline checklist with zero unaddressed items, and a final validated solution

Milestones

Week 2: Medallion lakehouse loads and transforms data end-to-end with reconciliation checks passing on all three layers

Week 4: Solution includes security, deployment promotion, orchestration, and a written architecture/runbook, with a diagnosed failure

Week 6: At least 3 workloads measurably improved, an automated monitoring element deployed, and alerting proven to fire

Week 8: Two full-length timed assessments at or above 70% in every domain and the skills-outline checklist fully closed out

Fabric DP-700 exam day

Your checklist

Government-issued photo ID

Bring the original, unexpired ID you registered with (passport or driver's license are typical). The first and last name on the ID must match your exam registration exactly - including spacing, hyphens, and middle names. If the names differ, contact the provider before test day to fix the registration.

Appointment confirmation and exam ID

Save a screenshot or offline copy of the confirmation email showing your appointment time, exam code (DP-700), time zone, and delivery method (test center vs. online proctored). You may need the registration/order number if you contact support.

Second form of ID (test-center candidates)

Some test centers request a second ID, often a credit card or employee badge with signature. Whether it is required varies by location - confirm with the provider when you book.

System and equipment check (online proctored)

Install or update the proctoring browser/app, then run the system test on the same machine, webcam, microphone, and network you will use on exam day. Do this at least 24 hours ahead so you have time to fix drivers or switch devices.

A private, quiet room with a clear desk

Remove papers, books, phones, smartwatches, second monitors, and clutter from the desk and surrounding area. Expect a room scan and possibly a desk-surface check before the exam begins.

Allowed scratch materials

Rules on physical notes, dry-erase sheets, or on-screen whiteboards differ by delivery method and proctor. Assume nothing is allowed unless the provider says so - confirm with the provider what note-taking is permitted.

Confirm allowed breaks and personal items

Policies on water in a clear container, tissues, medication, and restroom breaks vary by provider and test center. Confirm with the provider before test day rather than assuming.

Stable power and internet (online proctored)

Charge the laptop, have the charger plugged in, and prefer a wired connection or a strong signal. Tell household members you need an uninterrupted block of time.

Skills outline open the night before

Keep the official exam skills outline handy for a final skim: ingest and transform data, implement security and governance, monitor and optimize, and deploy/manage items. Use it as a map, not as a source of new material.

A short, personal weak-area sheet

Write a one-page summary in your own words of the topics you keep missing - for example Spark tuning knobs, OneLake security vs. workspace roles, deployment pipeline behavior, and Direct Lake semantics. Review this instead of re-reading everything.

Familiarity with the question formats

Expect multiple choice, drag-and-drop, hot-area, build-list, and case-study sets. Time-boxed case-study sections may not be revisitable once you leave them - confirm the behavior in the pre-exam tutorial on screen.

Backup plan for a technical failure

Know the provider's support contact before you start. If the session drops or the proctor chat freezes, note the time and reach out immediately rather than rebooting and hoping.

Plan the day

72 hours before

Re-read the skills outline and tag each objective as confident, shaky, or unknown. Spend your remaining time only on shaky and unknown items; do not start a brand-new topic this late.

72 hours before

Take the free official practice assessment once under timed conditions, then review every wrong answer and write down why it was wrong. Focus on reasoning patterns, not memorized answers.

72 hours before

Re-verify the appointment: date, start time, time zone, delivery method, and address or platform. Reschedule now if something is wrong - deadline and fee rules vary, so confirm with the provider.

72 hours before

Run the full system test if online proctored, and do a dry run of your room setup. Test the webcam, microphone, and headset now so a cable or driver issue does not surface on the day.

24 hours before

Do a light pass only: skim your weak-area sheet, re-read the outline headings, and stop studying early enough to sleep properly. Sleep is worth more than one more practice set.

24 hours before

Lay out your ID, charger, and confirmation details. Do not leave ID collection for the morning.

24 hours before

For a test center, confirm the route, parking or transit, and arrival lead time, and check for local disruptions. For online proctoring, confirm the room will be quiet and private at your slot.

Morning of

Eat normally, hydrate moderately, and avoid a heavy meal. Skip last-minute cramming of new material; one more unfamiliar fact usually adds anxiety, not points.

Morning of

Restart the computer, close every background app (chat, sync clients, email, VPN unless required), and disable notifications. Reconnect the charger before you launch the proctoring app.

Morning of

Log in or arrive earlier than the provider's stated lead time. Late arrivals can be refused entry or charged for a new appointment, so treat the lead time as a hard deadline - confirm the exact number with the provider.

During the test

Use the pre-exam tutorial screens (navigation, mark-for-review, comment box) fully before the timer starts. Then skim the total question count and set a rough per-question budget - around one to two minutes each, faster on short items.

During the test

Answer in two passes. Pass one: clear the questions you know and mark anything uncertain. Pass two: return to marked items and eliminate options rather than hunting for certainty.

During the test

Read each stem for the qualifier - 'first', 'least', 'most cost-effective', 'minimal effort'. In Fabric scenarios, eliminate options that require unsupported features, duplicate data unnecessarily, or ignore governance constraints.

During the test

Treat case studies as a separate unit. Read the business requirements first, then the technical requirements, and keep the requirements in mind for every item in that set. Once you exit a timed case-study section you may not be able to return - confirm the rule shown on screen.

During the test

Watch the clock at the halfway point and at ten minutes remaining. Leave a couple of minutes to confirm every question has a response; unanswered items score zero and there is no penalty for guessing.

After the test

Note any technical issues immediately and ask the proctor or support how to file a report before you close the session. Review your score report to see which skill areas were weak if you need to retake - retake waiting periods and fees are provider-specific, so confirm with them.

Rules to check

  • Identification: carry a valid, unexpired, government-issued photo ID whose name matches your registration exactly. A mismatch can mean being turned away and losing the fee, so fix discrepancies with the provider before exam day.

  • Authorized materials: assume nothing is permitted in the testing environment - no notes, printed material, phones, smartwatches, or additional screens. Any permitted aids (for example an on-screen whiteboard or dry-erase sheet) are determined by the provider, so confirm with the provider in advance.

  • Environment and proctoring: online sessions typically include an identity check, a room scan, and live or recorded monitoring. Follow the proctor's instructions exactly; refusing an inspection or leaving the camera view can end the session.

  • Breaks: scheduled and unscheduled break rules vary by delivery method, and the clock may keep running during a break. Confirm the break policy with the provider before your appointment and plan hydration around it.

  • Technical failures: if the session drops or the environment misbehaves, stop and contact the proctor or provider support right away and record the time. Do not quietly restart and continue, or you may lose the chance to have the attempt reinstated.

  • Scoring: Microsoft role-based exams are generally reported on a scaled score with 700 needed to pass, but the number of questions, time limit, and any section weighting can change. Verify the current details on the official exam page rather than relying on secondhand reports.

  • Confidentiality: you agree not to disclose, reproduce, or discuss exam content. Do not post questions, screenshots, or reconstructed scenarios anywhere, including study groups.

  • Rescheduling, cancellation, and retakes: notice periods, fees, and retake waiting rules are set by the provider and can change. Confirm with the provider instead of assuming a deadline, and check the policy before you make any change.

Fabric DP-700 FAQ

How many Fabric DP-700 practice questions are available?

The practice set has 200 Fabric DP-700 questions. Each one includes the choices, the correct answer, and an explanation, and the set is free to read with no signup.

How many Fabric DP-700 flashcards are available?

The flashcard set has 33 Fabric DP-700 cards. Each card shows the term and its definition.

Which topics does the Fabric DP-700 study material cover?

The Fabric DP-700 study material covers Implement and manage an analytics solution, Ingest and transform data, and Monitor and optimize an analytics solution.

What format facts does the catalog list for the Fabric DP-700 exam?

The catalog lists a duration of 100 minutes and these item types: proctored exam and may include interactive components for the Fabric DP-700 exam.