Compact MetricName tag map implementation #1368
Merged
Conversation
This file contains 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
Generate changelog in
|
carterkozak
commented
Feb 28, 2022
| static final TagMap EMPTY = new TagMap(new String[0]); | ||
|
|
||
| private final String[] values; | ||
| private final int hash; |
There was a problem hiding this comment.
I went back and forth on pre-computing the hash since the MetricName implementation already caches the hashCode.
It would be helpful for map wrappers with additional entries, however that case may perform better if we flatten it an create a new TagMap instance with an array two entries larger (assuming relatively few entries, which is generally true). Note that I have not verified this, so take it with a grain of salt.
In the common TaggedMetricSet case which applies additional entries, we never hash the name. It is passed to the metric log subscriber, and only hashed for gauges to ensure we're not blocked recording the same gauge twice.
There was a problem hiding this comment.
yeah, we could drop the hash field as all the current usages of TagMap will now compute the hashCode after creating a TagMap, and we now do create a new TagMap when adding an entry in https://github.com/palantir/tritium/pull/1368/files#diff-07d251c64776a9fd9112f154669063beff3ea2c65f4bece8c4bf9c24acce747aR87
Dropped the hash field from #1367
There was a problem hiding this comment.
Dropped hash field here as well
|
I'm preparing a change to remove the ExtraEntryMap wrapper in favor of using TagMap directly |
schlosna
reviewed
Mar 1, 2022
tritium-registry/src/main/java/com/palantir/tritium/metrics/registry/TagMap.java
Outdated
Show resolved
Hide resolved
schlosna
approved these changes
Mar 1, 2022
NestedMetricsBenchmark resultsBenchmarks after: Benchmarks before: AnalysisThese don't seem to tell the whole story, they measure object creation, but we never consume the metricname components, there's no iteration over keys. More representative benchmarksWith the benchmark changes introduced in 173efe9 we have larger tag maps, and the consumer blackholes the safeName and all tag key/value pairs. After:tags forEach(biconsumer)tags entrySet + enhanced for-loopBefore:tags forEach(biconsumer)tags entrySet + enhanced for-loop |
|
I was curious if the Adding this to the consumer: blackhole.consume(name.hashCode());Resulted in: The optimization gives us a 10% uplift, which is a very small faction of the improvement from this PR |
|
👍 |
|
Released 0.41.0 |
carterkozak
changed the title
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Before this PR
Excessive cost to create and store MetricName due to tags.
After this PR
==COMMIT_MSG==
Compact MetricName tag map implementation
==COMMIT_MSG==
Possible downsides?
Custom code