You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@kudu.apache.org by "Attila Bukor (Jira)" <ji...@apache.org> on 2019/11/21 00:00:00 UTC

[jira] [Assigned] (KUDU-33) Try to remove Schema information from logged writes

     [ https://issues.apache.org/jira/browse/KUDU-33?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Attila Bukor reassigned KUDU-33:
--------------------------------

    Assignee: Attila Bukor

> Try to remove Schema information from logged writes
> ---------------------------------------------------
>
>                 Key: KUDU-33
>                 URL: https://issues.apache.org/jira/browse/KUDU-33
>             Project: Kudu
>          Issue Type: Improvement
>          Components: perf, tablet, tserver
>    Affects Versions: M3
>            Reporter: Todd Lipcon
>            Assignee: Attila Bukor
>            Priority: Minor
>
> Currently the log records are the whole client request, including the schema information, etc. This can make them twice as big as they need to be, and might affect throughput. We should consider whether it's better to do all the decoding/schema projection "up front" on the leader, and then replicate a more "internal" form of the mutation, rather than replicating the client-provided request.



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