imay commented on issue #3288: Fix bug when use ZoneMap/BloomFiter on column with REPLACE/REPLACE_IF_NOT_NULL URL: https://github.com/apache/incubator-doris/pull/3288#issuecomment-611468986 > > If this optimization is removed, some query will suffer a performance degradation. > > When the scanned value column type is all REPLACE, it seems that this optimization can be applied. > > Can you leave this optimization for this case? > > I think it's a very corner case to select all columns on REPLACE and will waste much resource. > > 1. Only base rowset(version starts with zero) can use it, all other rowsets have to save ZoneMap/BloomFilter for REPLACE columns, which is non-sense. > 2. REPLACE is value column, the ZoneMap is not very useful because of sorted column is key column. Beside ZoneMap/BloomFilter, input filter will be applied before rowset streams are merged, this will cause this problem too. And in most cases, base rowset has the most rows, and filtering data before merging will reduce merge operation dramatically. It will result in a good performance. So I think ZoneMap and BloomFilter is not useful for REPLACE column, however filtering before merging worths it.
---------------------------------------------------------------- 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: us...@infra.apache.org With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org For additional commands, e-mail: commits-h...@doris.apache.org
