Question 520
For a public-sector reporting unit, this issue must not recur: A data app needs to ingest events from Kafka but engineers build one file per message and a new warehouse COPY for each row. Which review control fits best?
Explanation
Correct Answer
Benchmark actual event rate and failure recovery when selecting ingestion architecture. This reduces recurrence of: Connector-based streaming ingestion can be a better fit than per-event staged-file bulk COPY. The verification should demonstrate: The pipeline loads representative events with measured latency and manageable operational overhead.
Incorrect Answers
Test default role, warehouse and database values in deployment pipelines. That safeguard is aimed at ANOTHER risk: An application connects with the correct account but receives a warehouse-not-found error because it selects an unapproved default warehouse. It is less targeted than the control for: Connector-based streaming ingestion can be a better fit than per-event staged-file bulk COPY.
Keep environment-specific account identifiers out of source-code assumptions. That safeguard is aimed at ANOTHER risk: A Snowflake Python Connector job reaches the wrong account because the application uses a short legacy account locator. It is less targeted than the control for: Connector-based streaming ingestion can be a better fit than per-event staged-file bulk COPY.
Test schema drift and reordered source files before each upstream release. That safeguard is aimed at ANOTHER risk: An operations analyst assumes COPY INTO will map reordered source columns correctly solely because target columns have matching names. It is less targeted than the control for: Connector-based streaming ingestion can be a better fit than per-event staged-file bulk COPY.