DP-700 Exam Questions and Answers

Deployment Pipelines with Deployment Rules

Deployment pipelines facilitate the lifecycle management of Fabric items by allowing you to define rules that modify data source connections or parameters during deployment. This ensures that as items move from Test to Production, they automatically connect to the intended production databases, maintaining environment consistency and operational security.

The Fabric tenant setting 'Users can export items' must be enabled.

Git integration relies on the ability to export item metadata into a format compatible with source control. If the tenant setting for exporting items is disabled, Fabric cannot generate the necessary JSON representations of the items, effectively blocking any synchronization or connection to an Azure DevOps repository.

The Azure DevOps repository must be initialized with at least one file.

Fabric requires an initialized branch to establish a connection. An empty repository without a master or main branch cannot be targeted for synchronization. Initializing the repository with a README or gitignore file creates the initial commit necessary for Fabric to map the workspace structure to the branch.

Add a 'retryIntervalInSeconds' property to the retryPolicy.

Adding a retry interval allows the system to wait between attempts. If the failure was caused by a transient capacity spike or reaching the throttling limit, a delay gives the capacity time to recover or for other jobs to complete, significantly improving the success rate of subsequent attempts.

Implement Domains and assign workspaces to them.

Domains are designed exactly for this scenario. They allow you to group workspaces by department and assign Domain Contributors or Admins. This delegates the ability to manage those specific workspaces to the departments themselves while ensuring they still adhere to the broader tenant-level governance and security policies.

Data source rules

Data source rules allow you to redirect connections from a test database to a production database. When the pipeline runs, Fabric automatically replaces the connection details in the Lakehouse or semantic model metadata, ensuring that the production environment always processes the correct, live business data instead of test datasets.

Parameter rules

Many Fabric items use parameters to define variables like folder paths, dates, or thresholds. Parameter rules allow you to override these values during deployment. For instance, you can set a 'FilePath' parameter to point to a production container in Azure Data Lake Storage when moving from Test to Production.

Azure Key Vault secret rules

While often categorized under data sources or parameters, Fabric supports specific overrides for credentials and secrets. Ensuring that the Production stage uses a different secret from Key Vault than the Test stage is a critical security practice supported by the deployment engine to protect sensitive connection information.

Create a OneLake data access role for the specific folder.

OneLake data access roles allow you to define specific permissions (like Read) for individual folders within a Lakehouse. You can then assign the team to this role without giving them workspace-level permissions, ensuring they can only access the data in that specific directory via OneLake shortcuts or APIs.

Configure a Self-hosted Integration Runtime and enable the Staged Copy feature using an Azure Blob storage account.

Staged copy allows the ingestion process to buffer data in intermediate storage, which significantly boosts throughput. By offloading the data to an intermediary storage layer, the source system is freed up faster, and the Fabric pipeline can ingest the data in parallel, effectively scaling the throughput for large-scale operations.

The sink type is set to DeltaSink, which requires a pre-defined schema mapping for autoCreate.

Delta tables are strongly typed. When autoCreate is enabled in a copy activity, the engine needs an explicit schema definition to map the source file columns to appropriate Delta data types. Failing to provide this mapping prevents the creation of the underlying table with correct data types.

Fabric Dataflow Gen2.

Dataflow Gen2 offers a user-friendly Power Query interface that generates optimized Spark code for transformations. It allows for complex data cleansing, merging, and shaping without requiring deep coding expertise, making it perfect for developers who need to implement business logic across multiple disparate tables efficiently within Fabric.

OPTIMIZE table_name

The OPTIMIZE command directly addresses small file fragmentation by rewriting the data into larger, more efficient files. This process reduces metadata overhead and improves overall scan speeds, making it the correct solution for tables suffering from performance degradation due to a high volume of tiny data files.

If-Condition activity

The If-Condition activity allows developers to execute different sets of activities based on a logical true/false evaluation of an expression. This is the primary mechanism for implementing binary decision paths within a pipeline, such as checking if a source file exists before starting an ingestion task.

Switch activity

The Switch activity enables the pipeline to evaluate a value against multiple cases, executing distinct activities for each match. It is highly effective for controlling complex workflows where different source types or environments require specialized handling, providing a clean alternative to chaining multiple nested If-Condition activities together.

Use a Web activity to fetch the token, then pass it to the Copy activity.

This approach automates the credential lifecycle. By dynamically fetching a fresh token via a Web activity, the pipeline ensures that every run is authenticated properly. Storing the output of the Web activity as a variable allows it to be injected into the Copy activity headers securely.

The total number of files within the table directory.

Delta tables perform best when data is stored in optimally sized files. If a table contains thousands of small files, the query engine spends excessive time on metadata listing and initialization. Monitoring file count helps identify when the table requires a compaction process to merge these small files.

Capacity Utilization percentage

Capacity utilization represents the actual compute usage against your SKU. Tracking this metric allows you to visualize how smoothing impacts the overall load, showing you if your workloads are consistently peaking or if they remain within the sustained performance thresholds set by your current capacity tier.

Throttling Events

Throttling events occur when the smoothing mechanism cannot fully compensate for sustained demand exceeding the capacity limits. Monitoring these events is essential to determine if your current SKU is under-provisioned, as it directly indicates where smoothing has failed to prevent performance degradation for the users involved.

Increase the executor memory configuration for the Spark pool.

Z-ORDER operations require significant memory for sorting and shuffling data to align it by the specified columns. Increasing the executor memory provides the Spark engine with sufficient capacity to perform these intensive operations without exceeding the memory limits allocated to each node in the cluster.

Spark UI

The Spark UI provides an in-depth view of job execution, including DAG visualizations, stage durations, and task-level metrics. It is the primary tool for investigating performance bottlenecks within Spark jobs, allowing developers to drill down into why specific operations are taking longer than expected during execution.

Apply salting to the skewed join key.

Salting involves adding a random prefix to the join key, which forces the skewed data to be redistributed across multiple partitions. This prevents a single executor from bearing the entire burden of the skewed key, balancing the processing load more evenly across the entire compute cluster.

Enable Skewed Join Optimization in Spark configuration.

Spark has built-in mechanisms to detect and handle skewed joins automatically. By enabling this configuration, the engine intelligently splits skewed tasks into smaller sub-tasks, processing them in parallel on multiple executors, which significantly reduces the performance impact of uneven data distribution on specific join keys.

Pre-aggregate the skewed dataset.

Reducing the volume of the skewed dataset before joining it with other data is highly effective. By aggregating the skewed table, you shrink the amount of data the join operation needs to process, which naturally minimizes the severity of any existing skew that might have been present.

OPTIMIZE TABLE

The OPTIMIZE command is specifically designed to compact small Delta files into larger files. This process significantly improves read performance by reducing the metadata volume and enabling more efficient disk I/O, which is essential for maintaining query speed in tables that undergo frequent updates or high-frequency ingestion.