Use lossy summation for time-series aggregations #132625
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
dnhatn
force-pushed
the
sum-aggregator
branch
3 times, most recently
from
beb2b80 to
7c0d9f6
Compare
dnhatn
changed the title
Member
Author
{
"operator": "TimeSeriesAggregationOperator[blockHash=BytesRefLongBlockHash{keys=[BytesRefKey[channel=3], LongKey[channel=2]], entries=546, size=56368b}, aggregators=[GroupingAggregator[aggregatorFunction=SumDoubleGroupingAggregatorFunction[channels=[4]], mode=INITIAL], GroupingAggregator[aggregatorFunction=CountGroupingAggregatorFunction[channels=[4]], mode=INITIAL], GroupingAggregator[aggregatorFunction=ValuesBytesRefGroupingAggregatorFunction[channels=[5]], mode=INITIAL]]]",
"status": {
"hash_nanos": 2949462,
"aggregation_nanos": 22679014, // <- 22ms
"pages_processed": 546,
"rows_received": 982982,
"rows_emitted": 546,
"emit_nanos": 121951
}
}{
"operator": "TimeSeriesAggregationOperator[blockHash=BytesRefLongBlockHash{keys=[BytesRefKey[channel=3], LongKey[channel=2]], entries=546, size=56368b}, aggregators=[GroupingAggregator[aggregatorFunction=LossySumDoubleGroupingAggregatorFunction[channels=[4]], mode=INITIAL], GroupingAggregator[aggregatorFunction=CountGroupingAggregatorFunction[channels=[4]], mode=INITIAL], GroupingAggregator[aggregatorFunction=ValuesBytesRefGroupingAggregatorFunction[channels=[5]], mode=INITIAL]]]",
"status": {
"hash_nanos": 2770991,
"aggregation_nanos": 15664657, // <- 15ms
"pages_processed": 546,
"rows_received": 982982,
"rows_emitted": 546,
"emit_nanos": 72400
}
} |
28 tasks
dnhatn
added
:StorageEngine/TSDB
martijnvg
reviewed
Aug 11, 2025
Member
martijnvg
left a comment
martijnvg
left a comment
There was a problem hiding this comment.
Thanks Nhat, this is a great speed up for the TS command!
I left a question, but other than that this looks good to me.
| ) Expression field | ||
| ) { | ||
| this(source, field, Literal.TRUE); | ||
| this(source, field, Literal.TRUE, SummationMode.COMPENSATED_LITERAL); |
Member
There was a problem hiding this comment.
Should we consider also using LOSSY_LITERAL for avg the function is used in the context of TS source command? If so, then maybe we can do that in a followup?
Member
Author
There was a problem hiding this comment.
++. I will do it in a follow-up, as these may require bigger changes.
| ) | ||
| public Sum(Source source, @Param(name = "number", type = { "aggregate_metric_double", "double", "integer", "long" }) Expression field) { | ||
| this(source, field, Literal.TRUE); | ||
| this(source, field, Literal.TRUE, SummationMode.COMPENSATED_LITERAL); |
Member
There was a problem hiding this comment.
Same question as for the avg function.
Collaborator
|
Pinging @elastic/es-analytical-engine (Team:Analytics) |
elasticsearchmachine
added
Team:Analytics
Collaborator
|
Pinging @elastic/es-storage-engine (Team:StorageEngine) |
kkrik-es
reviewed
Aug 11, 2025
|
|
||
| public static void combineIntermediate(SumState state, double inValue, double zeroDelta, boolean seen) { | ||
| assert zeroDelta == 0.0 : zeroDelta; | ||
| if (seen) { |
Contributor
There was a problem hiding this comment.
Should this be seen == false ?
Member
Author
There was a problem hiding this comment.
Here, seen indicates that the input value is valid.
kkrik-es
reviewed
Aug 11, 2025
...ql/compute/src/main/java/org/elasticsearch/compute/aggregation/LossySumDoubleAggregator.java
Outdated
Show resolved
Hide resolved
kkrik-es
reviewed
Aug 11, 2025
| private static Avg readFrom(StreamInput in) throws IOException { | ||
| Source source = Source.readFrom((PlanStreamInput) in); | ||
| Expression field = in.readNamedWriteable(Expression.class); | ||
| Expression filter = in.getTransportVersion().onOrAfter(TransportVersions.V_8_16_0) |
Contributor
There was a problem hiding this comment.
Why do we use 8.16 here? Maybe add a comment?
Member
Author
There was a problem hiding this comment.
This was copied from the super class. I took a new approach in 8ec481f
Member
Author
|
Thanks friends! |
phananh1010
added a commit
to phananh1010/elasticsearch
that referenced
this pull request
Sep 23, 2025
BASE=30024e14edd361c4b9af134d6ddfc04ab1a061bc HEAD=8ec481f7a2bbe403cea5e2e52bb9985aa3473ac2 Branch=main
phananh1010
added a commit
to phananh1010/elasticsearch
that referenced
this pull request
Oct 2, 2025
BASE=30024e14edd361c4b9af134d6ddfc04ab1a061bc HEAD=8ec481f7a2bbe403cea5e2e52bb9985aa3473ac2 Branch=main
phananh1010
added a commit
to phananh1010/elasticsearch
that referenced
this pull request
Oct 7, 2025
BASE=30024e14edd361c4b9af134d6ddfc04ab1a061bc HEAD=8ec481f7a2bbe403cea5e2e52bb9985aa3473ac2 Branch=main
phananh1010
added a commit
to phananh1010/elasticsearch
that referenced
this pull request
Oct 17, 2025
BASE=30024e14edd361c4b9af134d6ddfc04ab1a061bc HEAD=8ec481f7a2bbe403cea5e2e52bb9985aa3473ac2 Branch=main
phananh1010
added a commit
to phananh1010/elasticsearch
that referenced
this pull request
Oct 23, 2025
BASE=30024e14edd361c4b9af134d6ddfc04ab1a061bc HEAD=8ec481f7a2bbe403cea5e2e52bb9985aa3473ac2 Branch=main
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Kahan summation can be expensive, and for time-series aggregation, a lossy summation can be a good trade-off for performance. This change introduces a lossy summation mode and makes it the default for time-series aggregations. These two summation modes for sum and avg are used internally and are not exposed to users.