Details
-
Epic
-
Status: Closed
-
Major
-
Resolution: Duplicate
-
None
-
None
-
0
Description
Hudi has indices to assit in tagging incoming records. Most commonly used one is Bloom index. This involves looking up (loading) bloom from data files which could be time consuming and could have throttling impact in cloud stores like S3. So, proposing this RFC to add bloom as a special partition in metadata table and implement an index based on that.
Attachments
Issue Links
- duplicates
-
HUDI-1822 [Umbrella] Multi Modal Indexing
- Closed