You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We have a use-case that requires high-qps against an Upsert table. While looking into the bottlenecks we found that the majority of allocations (68.44%) are caused by the bitmap cloning done by the ThreadSafeMutableRoaringBitmap.
I am assuming #13677 in Pinot 1.3 fixes this. Can someone confirm? Also can someone update Pinot Docs with details on the consistency view feature now that it's about to be released with 1.3? cc: @klsince
The text was updated successfully, but these errors were encountered:
Not particularly made for solve this bottleneck, but the ‘snapshot’ mode may help here, as IIRC the query could reuse the bitmap copies as tracked by the snapshot view w/o doing clone.
We have a use-case that requires high-qps against an Upsert table. While looking into the bottlenecks we found that the majority of allocations (68.44%) are caused by the bitmap cloning done by the
ThreadSafeMutableRoaringBitmap
.I am assuming #13677 in Pinot 1.3 fixes this. Can someone confirm? Also can someone update Pinot Docs with details on the consistency view feature now that it's about to be released with 1.3? cc: @klsince
The text was updated successfully, but these errors were encountered: