Configuring a retry policy is the native, built-in mechanism in OCI Data Integration for handling transient failures. It allows the pipeline to automatically re-attempt execution after a configurable delay, which is ideal for scenarios like temporary source unavailability during maintenance. This approach requires minimal custom logic and leverages the service’s reliability features.

Conditional links (or branches) in an OCI Data Integration pipeline allow you to control the flow of execution based on the outcome of a previous task. You can configure an expression on the link (e.g., `$lastTask.output.rowCount > 1000`) to determine whether the subsequent task is executed. This is the direct, declarative way to build conditional workflow logic within the pipeline itself.

Enabling partitioning for the write operation to ADW allows OCI Data Integration to parallelize the load. It splits the data based on the partition key and writes multiple chunks concurrently into the target table, which can dramatically improve performance for large datasets. This leverages ADW’s ability to handle parallel data loads effectively.

The correct pattern is to define pipeline-level parameters. The output of the REST Start task (which includes the incoming payload) can be mapped to these parameters. Once bound, the parameter’s value becomes accessible throughout the rest of the pipeline, including within data flow expressions using the syntax `$parameter.parameterName`. This provides a clean, managed way to pass context between tasks.

The ‘Limit’ option in the data flow’s runtime configuration is specifically designed for this purpose. When you configure a limit (e.g., 1000 rows), the integration service will only fetch that many rows from the source(s), allowing for fast, iterative testing and debugging of transformation logic without the cost or time of processing a full dataset.

A ‘Function’ operator allows you to apply SQL-like functions to attributes. Placing it before the ‘Aggregate’ operator and using it to create a new output attribute with `UPPER(region)` will normalize all region values to uppercase. You would then use this new attribute as the group-by key in the subsequent ‘Aggregate’ operator, ensuring case-insensitive grouping.

The ‘Function’ operator is the primary tool for data type conversion within a data flow. You can use an expression like `TO_NUMBER(string_key_column)` to create a new attribute with the correct numeric data type. This new attribute can then be used as the join condition with the database table’s numeric key, resolving the type mismatch within the integration logic itself.

OCI Events Service provides native, event-driven integration for OCI services. OCI Data Integration emits lifecycle events, including ‘Pipeline Run Failed’. You can create a rule in OCI Events to capture this specific event and route it to an action, such as posting to an ONS (Oracle Notification Service) topic which can send emails, SMS, or trigger Functions. This is a serverless, managed solution that requires no polling or custom pipeline logic.

For high-volume sources, parallelism is key. The ‘Number of Partitions’ setting in the source operator determines how many chunks the source data is split into, enabling parallel reading. The ‘Driver Shape’ and ‘Executor Shape’ (though listed as one option in the answer choices, it’s a core concept) define the compute resources allocated to the integration runtime, directly impacting how many parallel tasks can be executed efficiently. ‘Batch Size’ affects commit frequency, not read parallelism, and ‘Update Frequency’ is a scheduling concern.

Configuring a retry policy is the native, built-in mechanism in OCI Data Integration for handling transient failures. It allows the pipeline to automatically re-attempt execution after a configurable delay, which is ideal for scenarios like temporary source unavailability during maintenance. This approach requires minimal custom logic and leverages the service’s reliability features.0