[jira] [Commented] (CASSANDRA-3262) SimpleSnitch.compareEndpoints doesn't respect the intent of the snitch

["Jonathan Ellis (Commented) (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CASSANDRA-3262?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13115458#comment-13115458 ] 

Jonathan Ellis commented on CASSANDRA-3262:
-------------------------------------------

The reason for this behavior (CASSANDRA-1314) is because for best performance, normal reads want exactly the opposite behavior of counters: we want to direct reads to the same replica so that the cache stays hot.  Put another way, if each replica is serving a distinct range then you get (replica count) times as much cache memory [with CL.ONE and RR off) than if each is getting the full range of requests from different coordinators.
                
> SimpleSnitch.compareEndpoints doesn't respect the intent of the snitch
> ----------------------------------------------------------------------
>
>                 Key: CASSANDRA-3262
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-3262
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>    Affects Versions: 0.8.6
>            Reporter: Sylvain Lebresne
>            Assignee: Sylvain Lebresne
>            Priority: Minor
>             Fix For: 0.8.7, 1.0.0
>
>         Attachments: 3262.patch
>
>
> SimpleSnitch is supposed to not sort the input addresses, thus respecting the order of the partitioner. However, it's compareEndpoints instead uses IP addresses comparison. Note that this matter when the dynamicSnitch fall back to the wrapped snitch since it uses the compareEndpoint method then.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira