siddharthteotia opened a new pull request #5013: For RANGE predicate queries touching offline segments, use sorted inverted index. URL: https://github.com/apache/incubator-pinot/pull/5013 Motivation: Currently, we don't leverage inverted index (sorted or bitmap) for RANGE predicates. Rather we resort to scan based iterator. If a column is sorted and has RANGE predicates like: WHERE COL >= 20000 WHERE COL >= 20000 AND COL <= 25000 ..... Using sorted inverted index is better in such cases since the predicate evaluator can quickly give us the matching dictionary IDs, these are already sorted. All we need now is get the startDocId for dictid[0] and endDocId for dictId[dictids.length - 1]. This change is made in SortedInvertedIndexBasedFilterOperator. Later we can iterate on these or intersect. We can only do this for offline segments since for realtime segments, until the conversion happens the records aren't really sorted.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
