[
https://issues.apache.org/jira/browse/KAFKA-10470?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17218064#comment-17218064
]
Chia-Ping Tsai commented on KAFKA-10470:
----------------------------------------
[~yuzawa-san] Thanks for your effort on zstd-jni. Do you want to take over this
issue to leverage the new BufferPool you introduced ?
There are two tasks for this issue.
1. update zstd to v1.4.5-8 to use the new interface (BufferPool) introduced by
your PR (https://github.com/luben/zstd-jni/pull/146)
2. implement the BufferPool of zstd to reuse the byte buffer
(https://github.com/apache/kafka/blob/trunk/clients/src/main/java/org/apache/kafka/common/record/CompressionType.java)
Also, there is a PR (https://github.com/apache/kafka/pull/9229) trying to use
true reusable BufferSupplier to create IO stream (before the PR, the passed
BufferSupplier does not recycle the byte buffer).
> zstd decompression with small batches is slow and causes excessive GC
> ---------------------------------------------------------------------
>
> Key: KAFKA-10470
> URL: https://issues.apache.org/jira/browse/KAFKA-10470
> Project: Kafka
> Issue Type: Bug
> Affects Versions: 2.5.1
> Reporter: Robert Wagner
> Priority: Major
>
> Similar to KAFKA-5150 but for zstd instead of LZ4, it appears that a large
> decompression buffer (128kb) created by zstd-jni per batch is causing a
> significant performance bottleneck.
> The next upcoming version of zstd-jni (1.4.5-7) will have a new constructor
> for ZstdInputStream that allows the client to pass its own buffer. A similar
> fix as [PR #2967|https://github.com/apache/kafka/pull/2967] could be used to
> have the ZstdConstructor use a BufferSupplier to re-use the decompression
> buffer.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
