[GitHub] [incubator-pinot] mayankshriv commented on a change in pull request #5419: [WIP]Use thread local for groupby raw key holders

[GitBox <gi...@apache.org>]

mayankshriv commented on a change in pull request #5419:
URL: https://github.com/apache/incubator-pinot/pull/5419#discussion_r428218159



##########
File path: pinot-core/src/main/java/org/apache/pinot/core/query/aggregation/groupby/DictionaryBasedGroupKeyGenerator.java
##########
@@ -106,18 +108,26 @@ public DictionaryBasedGroupKeyGenerator(TransformOperator transformOperator,
 
       _isSingleValueColumn[i] = transformOperator.getResultMetadata(groupByExpression).isSingleValue();
     }
-
     if (longOverflow) {
       _globalGroupIdUpperBound = numGroupsLimit;
-      _rawKeyHolder = new ArrayMapBasedHolder(_globalGroupIdUpperBound);
+      if (!mapBasedRawKeyHolders.containsKey(ArrayMapBasedHolder.class.getName())) {
+        mapBasedRawKeyHolders.put(ArrayMapBasedHolder.class.getName(), new ArrayMapBasedHolder(_globalGroupIdUpperBound).getInternal());

Review comment:
       I think initializing to `_globalGroupIdUpperBound` got introduced in #5291. For many cases with multiple group by columns (high cardinality and/or MV columns) this number can be huge. Unclear to me if making this thread-local will protect against such cases that may require allocating huge chunk of memory upfornt.




----------------------------------------------------------------
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:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@pinot.apache.org
For additional commands, e-mail: commits-help@pinot.apache.org