Summary

The existing QueriesPipelineExecutor uses a block-based scheduling model. This granularity is too coarse for effective resource management and fine-grained control.

This tracking issue outlines the work required to enhance the executor to a production-ready state by implementing a more precise, row-based scheduling system.

• refactor(executor): control queries executor via configuration #18283

• feat(executor): add query execution stats #18345 (Implement a new metric to count how many rows are scheduled for each query_id during a epoch)

• Improve the current queries scheduler by switching to a row-level scheduling approach

• Combine with the workload group feature to enable more accurate and detailed CPU resource management.