Runtime filtering reduces data movement in federated query execution by deriving relevant values from one side of a join and applying them to the larger dataset before retrieval. This lowers the data scanned, transmitted, and processed, improving latency and resource use. section 8.3.4
Runtime filtering is most valuable when:
- A small table identifies a limited set of values needed from a much larger remote table. section 8.3.4
- Federated queries would otherwise transfer substantial data across remote systems or network infrastructure. section 8.3.4
- A semantic layer supplies relationship and join logic that helps locate effective filter points. section 8.3.4
Filtering a large joined dataset
A federated query first identifies relevant values from a small table, then applies those values while retrieving the larger joined dataset, avoiding unnecessary scans and transfers. section 8.3.4
