You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@druid.apache.org by GitBox <gi...@apache.org> on 2021/11/23 10:40:17 UTC
[GitHub] [druid] kfaraz commented on a change in pull request #11950: Improve on-heap aggregator footprint estimates.
kfaraz commented on a change in pull request #11950:
URL: https://github.com/apache/druid/pull/11950#discussion_r754990694
##########
File path: processing/src/main/java/org/apache/druid/segment/incremental/OnheapIncrementalIndex.java
##########
@@ -114,10 +122,13 @@
private static long getMaxBytesPerRowForAggregators(IncrementalIndexSchema incrementalIndexSchema)
{
long maxAggregatorIntermediateSize = ((long) Integer.BYTES) * incrementalIndexSchema.getMetrics().length;
- maxAggregatorIntermediateSize += Arrays.stream(incrementalIndexSchema.getMetrics())
- .mapToLong(aggregator -> aggregator.getMaxIntermediateSizeWithNulls()
- + Long.BYTES * 2L)
- .sum();
+
+ for (final AggregatorFactory aggregator : incrementalIndexSchema.getMetrics()) {
+ final long rows =
Review comment:
Nit: It seems this line can be outside the for loop.
##########
File path: extensions-core/datasketches/src/main/java/org/apache/druid/query/aggregation/datasketches/theta/SketchAggregatorFactory.java
##########
@@ -170,6 +171,25 @@ public int getSize()
return size;
}
+ @Override
+ public int guessAggregatorHeapFootprint(long rows)
+ {
+ final int maxEntries = size * 2;
+ final int expectedEntries;
+
+ if (rows > maxEntries) {
+ expectedEntries = maxEntries;
+ } else {
+ final int minEntries = 1 << Util.MIN_LG_ARR_LONGS;
Review comment:
Should `minEntries = 1 << Util.MIN_LG_ARR_LONGS` and the largest possible preamble, `Family.UNION.getMaxPreLongs() << 3` be taken out into constants?
##########
File path: processing/src/main/java/org/apache/druid/segment/incremental/OnheapIncrementalIndex.java
##########
@@ -114,10 +122,13 @@
private static long getMaxBytesPerRowForAggregators(IncrementalIndexSchema incrementalIndexSchema)
{
long maxAggregatorIntermediateSize = ((long) Integer.BYTES) * incrementalIndexSchema.getMetrics().length;
- maxAggregatorIntermediateSize += Arrays.stream(incrementalIndexSchema.getMetrics())
- .mapToLong(aggregator -> aggregator.getMaxIntermediateSizeWithNulls()
- + Long.BYTES * 2L)
- .sum();
+
+ for (final AggregatorFactory aggregator : incrementalIndexSchema.getMetrics()) {
+ final long rows =
+ incrementalIndexSchema.isRollup() ? ROLLUP_RATIO_FOR_AGGREGATOR_FOOTPRINT_ESTIMATION : 1;
+ maxAggregatorIntermediateSize += (long) aggregator.guessAggregatorHeapFootprint(rows) + 2 * Long.BYTES;
Review comment:
Nit: Is the term `2 * Long.BYTES (= 16)` needed now?
It seems that that the result of the guess would be big enough even when `rows = 1` to make this term relatively much less significant.
--
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.
To unsubscribe, e-mail: commits-unsubscribe@druid.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@druid.apache.org
For additional commands, e-mail: commits-help@druid.apache.org