Use buffered writer #1241
Merged
Use buffered writer #1241
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
Without buffered reader, every single output character through given stack goes to "StreamEncoder public void write(int c)" and single element char array is created.
It makes pressure to GC and CPU
BufferedReader reduce the pressure
write:130, StreamEncoder (sun.nio.cs)
write:201, OutputStreamWriter (java.io)
writeMetadata:338, PrometheusTextFormatWriter (io.prometheus.metrics.expositionformats)
writeGauge:129, PrometheusTextFormatWriter (io.prometheus.metrics.expositionformats)
write:68, PrometheusTextFormatWriter (io.prometheus.metrics.expositionformats)
write:48, PrometheusOutputFormat$1 (org.springframework.boot.actuate.metrics.export.prometheus)
scrape:87, PrometheusScrapeEndpoint (org.springframework.boot.actuate.metrics.export.prometheus)
StreamEncoder public void write(int c) throws IOException {
char[] cbuf = new char[1];
cbuf[0] = (char) c;
Signed-off-by: Michal Domagala <michaldo@github.io>
michaldo
force-pushed
the
use-buffered-writer
branch
from
b26dd34 to
b9421b5
Compare
|
Thanks for the contribution! |
Merged
This was referenced Feb 15, 2025
Open
dhoard
added a commit
that referenced
this pull request
Feb 18, 2025
Further optimize `TextFormatUtil#writeEscapedLabelValue`. We're seeing `TextFormatUtil#writeEscapedLabelValue` show up in our production traces due to the single `char` writes to `OutputStreamWriter`. We're using `OpenMetricsTextFormatWriter`. #1241 and #1248 should take care of most of these issues but there still remains some optimization potential left. `BufferedWriter#write(int)` has some minimal overhead in terms of locking. If we assume that rarely any characters need to be escaped and instead optimize to write as large of a part of the String as possible in one method call. Before --------- ``` Benchmark Mode Cnt Score Error Units TextFormatUtilBenchmark.openMetricsWriteToByteArray thrpt 25 438485.372 ± 4270.355 ops/s TextFormatUtilBenchmark.openMetricsWriteToNull thrpt 25 440105.281 ± 2891.572 ops/s TextFormatUtilBenchmark.prometheusWriteToByteArray thrpt 25 467213.001 ± 878.780 ops/s TextFormatUtilBenchmark.prometheusWriteToNull thrpt 25 472931.759 ± 976.028 ops/s ``` After ------- ``` Benchmark Mode Cnt Score Error Units TextFormatUtilBenchmark.openMetricsWriteToByteArray thrpt 25 462852.243 ± 5071.696 ops/s TextFormatUtilBenchmark.openMetricsWriteToNull thrpt 25 469910.681 ± 1670.430 ops/s TextFormatUtilBenchmark.prometheusWriteToByteArray thrpt 25 482362.506 ± 2051.684 ops/s TextFormatUtilBenchmark.prometheusWriteToNull thrpt 25 487707.557 ± 3344.881 ops/s ``` About a 5% to 6% gain for `OpenMetricsTextFormatWriter` and around 3% for `PrometheusTextFormatWriter`. Note that this benchmark is actually for `OpenMetricsTextFormatWriter` and `PrometheusTextFormatWriter` since `TextFormatUtil` is not public and can therefore not be benchmarked directly. The relative gains in `#writeEscapedLabelValue` are higher because the benchmark runs the full text format writers. I also added a test for `TextFormatUtil#writeEscapedLabelValue` since the code is now more complicated. --------- Signed-off-by: Philippe Marschall <philippe.marschall@gmail.com> Co-authored-by: Doug Hoard <dhoard@users.noreply.github.com>
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.
Without buffered reader, every single output character through given stack goes to "StreamEncoder public void write(int c)" and single element char array is created. It makes pressure to GC and CPU
In my case 800K char[1] arrays are created and garbage collected
BufferedReader reduce the pressure