[
https://issues.apache.org/jira/browse/HIVE-22446?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16967803#comment-16967803
]
Slim Bouguerra commented on HIVE-22446:
---------------------------------------
[~t3rmin4t0r] can you take a look at this ? if you are okay will add more
testing
> Make IO decoding quantiles counters less contended resource.
> ------------------------------------------------------------
>
> Key: HIVE-22446
> URL: https://issues.apache.org/jira/browse/HIVE-22446
> Project: Hive
> Issue Type: Improvement
> Components: llap
> Reporter: Slim Bouguerra
> Assignee: Slim Bouguerra
> Priority: Major
> Fix For: 4.0.0
>
> Attachments: HIVE-22446.patch
>
>
> Currently LLAP IO relies on Hadoop Lock-based quantiles data structure and
> updates the IO decoding sample on a per batch based using.
> {code}
> org.apache.hadoop.hive.llap.metrics.LlapDaemonIOMetrics#addDecodeBatchTime
> {code}
> via
> {code}
> org.apache.hadoop.hive.llap.io.decode.EncodedDataConsumer#consumeData
> {code}
> This can be a source of thread contention.
> Goal of this ticket is to reduce the frequency of updates to avoid major
> bottleneck.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
