You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ozone.apache.org by "Marton Elek (Jira)" <ji...@apache.org> on 2020/02/21 14:20:00 UTC

[jira] [Updated] (HDDS-3053) Decrease the number of the chunk writer threads

     [ https://issues.apache.org/jira/browse/HDDS-3053?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Marton Elek updated HDDS-3053:
------------------------------
    Description: 
As of now we create 60 threads (
dfs.container.ratis.num.write.chunk.threads) to write chunk data to the disk. As the write is limited by the IO I can't see any benefit to have so many threads. High number of thread means a high context switch overhead, therefore it seems to be more reasonable to use only a limited number of threads.

For example 10 threads should be enough even with 5 external disk.

If you know any reason to keep the number 60, please let me know...


  was:
As of now we create 60 threads (
dfs.container.ratis.num.write.chunk.threads) to write chunk date to the disk. As the write is limited by the IO I can't see any benefit to have so many threads. High number of thread means a high context switch overhead, therefore it seems to be more reasonable to use only a limited number of threads.

For example 10 threads should be enough even with 5 external disk.

If you know any reason to keep the number 60, please let me know...



> Decrease the number of the chunk writer threads
> -----------------------------------------------
>
>                 Key: HDDS-3053
>                 URL: https://issues.apache.org/jira/browse/HDDS-3053
>             Project: Hadoop Distributed Data Store
>          Issue Type: Improvement
>            Reporter: Marton Elek
>            Priority: Major
>
> As of now we create 60 threads (
> dfs.container.ratis.num.write.chunk.threads) to write chunk data to the disk. As the write is limited by the IO I can't see any benefit to have so many threads. High number of thread means a high context switch overhead, therefore it seems to be more reasonable to use only a limited number of threads.
> For example 10 threads should be enough even with 5 external disk.
> If you know any reason to keep the number 60, please let me know...



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: ozone-issues-help@hadoop.apache.org