[
https://issues.apache.org/jira/browse/FLINK-14346?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16984917#comment-16984917
]
Roman Grebennikov commented on FLINK-14346:
-------------------------------------------
I've made a proper PR to port these performance optimizations from the PoC to
Flink codebase. So it would be really nice if someone will review them
eventually.
The motivation behind this change is listed in the PR description, but the most
important benchmark results looking quite good:
{noformat}
[info] Benchmark (length) (stringType) Mode
Cnt Score Error Units
[info] StringDeserializerBenchmark.deserializeDefault 1 ascii avgt
25 46.603 ± 0.750 ns/op
[info] StringDeserializerBenchmark.deserializeImproved 1 ascii avgt
25 51.074 ± 0.720 ns/op
[info] StringDeserializerBenchmark.deserializeJDK 1 ascii avgt
25 63.402 ± 1.631 ns/op
[info] StringSerializerBenchmark.serializeDefault 1 ascii avgt
25 31.595 ± 0.489 ns/op
[info] StringSerializerBenchmark.serializeImproved 1 ascii avgt
25 33.454 ± 0.151 ns/op
[info] StringSerializerBenchmark.serializeJDK 1 ascii avgt
25 34.721 ± 0.128 ns/op
[info] StringDeserializerBenchmark.deserializeDefault 16 ascii avgt
25 251.321 ± 3.251 ns/op
[info] StringDeserializerBenchmark.deserializeImproved 16 ascii avgt
25 55.385 ± 1.176 ns/op
[info] StringDeserializerBenchmark.deserializeJDK 16 ascii avgt
25 77.147 ± 1.661 ns/op
[info] StringSerializerBenchmark.serializeDefault 16 ascii avgt
25 95.782 ± 0.261 ns/op
[info] StringSerializerBenchmark.serializeImproved 16 ascii avgt
25 51.806 ± 0.180 ns/op
[info] StringSerializerBenchmark.serializeJDK 16 ascii avgt
25 50.786 ± 1.677 ns/op
[info] StringDeserializerBenchmark.deserializeDefault 128 ascii avgt
25 1757.726 ± 3.312 ns/op
[info] StringDeserializerBenchmark.deserializeImproved 128 ascii avgt
25 140.374 ± 1.006 ns/op
[info] StringDeserializerBenchmark.deserializeJDK 128 ascii avgt
25 263.445 ± 6.912 ns/op
[info] StringSerializerBenchmark.serializeDefault 128 ascii avgt
25 670.627 ± 2.807 ns/op
[info] StringSerializerBenchmark.serializeImproved 128 ascii avgt
25 161.481 ± 2.798 ns/op
[info] StringSerializerBenchmark.serializeJDK 128 ascii avgt
25 151.789 ± 7.295 ns/op
{noformat}
> Performance issue with StringSerializer
> ---------------------------------------
>
> Key: FLINK-14346
> URL: https://issues.apache.org/jira/browse/FLINK-14346
> Project: Flink
> Issue Type: Improvement
> Components: API / Type Serialization System, Benchmarks
> Affects Versions: 1.9.0
> Environment: Tested on Flink 1.9.0, adoptopenjdk 8u222.
> Reporter: Roman Grebennikov
> Priority: Major
> Labels: performance, pull-request-available
> Time Spent: 10m
> Remaining Estimate: 0h
>
> While doing a performance profiling for our Flink state-heavy streaming job,
> we found that quite a significant amount of CPU time is spent inside
> StringSerializer writing data to the underlying byte buffer. The hottest part
> of the code is the StringValue.writeString function. And replacing the
> default StringSerializer with the custom one (to just play with a baseline),
> which is just calling DataOutput.writeUTF/readUTF surprisingly yielded to
> almost 2x speedup for string serialization.
> As writeUTF and writeString have incompatible wire formats, replacing latter
> with former is not a good idea in general as it may break
> checkpoint/savepoint compatibility.
> We also did an early performance analysis of the root cause of this
> performance issue, and the main reason of JDK's writeUTF being faster is that
> it's code is not writing directly to output stream byte-by-byte, but instead
> creating an underlying temporary byte buffer. This yields to a HotSpot almost
> perfectly unrolling the main loop, which results in much better data
> parallelism.
> I've tried to port the ideas from the JVM's implementation of writeUTF back
> to StringValue.writeString, and my current result is nice, having quite
> significant speedup compared to the current implementation:
> {{[info] Benchmark Mode Cnt Score Error Units}}
> {{[info] StringSerializerBenchmark.measureJDK avgt 30 82.871 ± 1.293 ns/op}}
> {{[info] StringSerializerBenchmark.measureNew avgt 30 94.004 ± 1.491 ns/op}}
> {{[info] StringSerializerBenchmark.measureOld avgt 30 156.905 ± 3.596 ns/op}}
>
> {{Where measureJDK is the JDK's writeUTF asa baseline, measureOld is the
> current upstream implementation in Flink, and the measureNew is the improved
> one. }}
>
> {{The code for the benchmark (and the improved version of the serializer) is
> here: [https://github.com/shuttie/flink-string-serializer]}}
>
> {{Next steps:}}
> # {{More benchmarks for non-ascii strings.}}
> # {{Benchmarks for long strings.}}
> # {{Benchmarks for deserialization.}}
> # {{Tests for old-new wire format compatibility.}}
> # {{PR to the Flink codebase.}}
> {{Is there an interest for this kind of performance improvement?}}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
