You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "stack (JIRA)" <ji...@apache.org> on 2016/11/16 22:24:58 UTC

[jira] [Resolved] (HBASE-6773) Make the dfs replication factor configurable per table

     [ https://issues.apache.org/jira/browse/HBASE-6773?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

stack resolved HBASE-6773.
--------------------------
    Resolution: Invalid

No longer valid.

Can set by CF using the key HColumnDescriptor.DEFAULT_DFS_REPLICATION (It could be done in a nicer way but its doable).

> Make the dfs replication factor configurable per table
> ------------------------------------------------------
>
>                 Key: HBASE-6773
>                 URL: https://issues.apache.org/jira/browse/HBASE-6773
>             Project: HBase
>          Issue Type: Improvement
>          Components: regionserver
>    Affects Versions: 0.95.2
>            Reporter: Nicolas Liochon
>            Assignee: Devaraj Das
>
> Today, it's an application level configuration. So all the HFiles are replicated 3 times per default.
> There are some reasons to make it per table:
> - some tables are critical while some others are not. For example, meta would benefit from an higher level of replication, to ensure we continue working even when we lose 20% of the cluster.
> - some tables are backuped somewhere else, used by non essential process, so the user may accept a lower level of replication for these ones.
> - it should be a dynamic parameter. For example, during a bulk load we set a replication of 1 or 2, then we increase it. It's in the same space as disabling the WAL for some writes.
> The case that seems important to me is meta. We can also handle this one by a specific parameter in the usual hbase-site.xml if we don't want a generic solution.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)