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

Posted to issues@kudu.apache.org by "Jean-Daniel Cryans (JIRA)" <ji...@apache.org> on 2017/08/28 17:55:00 UTC

[jira] [Updated] (KUDU-1728) Parallelize tablet copy operations

     [ https://issues.apache.org/jira/browse/KUDU-1728?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jean-Daniel Cryans updated KUDU-1728:
-------------------------------------
    Target Version/s: Backlog

From recent experience, I think this would be great. If we could define a tserver-wide IO budget and on-the-fly decide if a tablet copy can do more (or less) in parallel then this could resolve some re-replication situations faster.

Another similar idea, maybe instead of doing many tablets in parallel we should instead focus on re-replicating single tablets as fast as we can? It might become more difficult to get good parallelism over a cluster though since there's fewer copy operations going on.

> Parallelize tablet copy operations
> ----------------------------------
>
>                 Key: KUDU-1728
>                 URL: https://issues.apache.org/jira/browse/KUDU-1728
>             Project: Kudu
>          Issue Type: Bug
>          Components: consensus, tablet
>            Reporter: Mike Percy
>
> Parallelize tablet copy operations. Right now all data is copied serially. We may want to consider throttling on either side if we want to budget IO.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)