kafka-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jason Gustafson (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (KAFKA-3704) Improve mechanism for compression stream block size selection in KafkaProducer
Date Tue, 27 Sep 2016 23:25:22 GMT

     [ https://issues.apache.org/jira/browse/KAFKA-3704?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Jason Gustafson updated KAFKA-3704:
    Fix Version/s:     (was:

> Improve mechanism for compression stream block size selection in KafkaProducer
> ------------------------------------------------------------------------------
>                 Key: KAFKA-3704
>                 URL: https://issues.apache.org/jira/browse/KAFKA-3704
>             Project: Kafka
>          Issue Type: Improvement
>            Reporter: Guozhang Wang
>            Assignee: Ismael Juma
>             Fix For:
> As discovered in https://issues.apache.org/jira/browse/KAFKA-3565, the current default
block size (1K) used in Snappy and GZIP may cause a sub-optimal compression ratio for Snappy,
and hence reduce throughput. Because we no longer recompress data in the broker, it also impacts
what gets stored on disk.
> A solution might be to use the default block size, which is 64K in LZ4, 32K in Snappy
and 0.5K in GZIP. The downside is that this solution will require more memory allocated outside
of the buffer pool and hence users may need to bump up their JVM heap size, especially for
MirrorMakers. Using Snappy as an example, it's an additional 2x32k per batch (as Snappy uses
two buffers) and one would expect at least one batch per partition. However, the number of
batches per partition can be much higher if the broker is slow to acknowledge producer requests
(depending on `buffer.memory`, `batch.size`, message size, etc.).
> Given the above, there are a few things that could be done (potentially more than one):
> 1) A configuration for the producer compression stream buffer size.
> 2) Allocate buffers from the buffer pool and pass them to the compression library. This
is possible with Snappy and we could adapt our LZ4 code. It's not possible with GZIP, but
it uses a very small buffer by default.
> 3) Close the existing `RecordBatch.records` when we create a new `RecordBatch` for the
`TopicPartition` instead of doing it during `RecordAccumulator.drain`. This would mean that
we would only retain resources for one `RecordBatch` per partition, which would improve the
worst case scenario significantly.
> Note that we decided that this change was too risky for and reverted the original

This message was sent by Atlassian JIRA

View raw message