[
https://issues.apache.org/jira/browse/CASSANDRA-839?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12839758#action_12839758
]
Gary Dusbabek commented on CASSANDRA-839:
-----------------------------------------
Even after 825 the mapping is still needed by the commit log.
Ideally, I'd like to get rid of the ids altogether. It seems they're only
really needed to keep the size of the CL header smaller. It makes more sense
to use some combination of KS name + CF name, or hashing on both if we're
really interested in compactness.
> Use hashes for cfids
> --------------------
>
> Key: CASSANDRA-839
> URL: https://issues.apache.org/jira/browse/CASSANDRA-839
> Project: Cassandra
> Issue Type: Sub-task
> Reporter: Gary Dusbabek
> Assignee: Gary Dusbabek
> Priority: Minor
> Fix For: 0.7
>
>
> Currently, cfids are generated at runtime using a simple counter. We need a
> more predictable approach to generating these ids so they can be guessed,
> rather than assigned.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
