You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Jonathan Ellis (JIRA)" <ji...@apache.org> on 2009/12/02 00:27:20 UTC
[jira] Resolved: (CASSANDRA-476) Pluggable Load and contrib example
[ https://issues.apache.org/jira/browse/CASSANDRA-476?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jonathan Ellis resolved CASSANDRA-476.
--------------------------------------
Resolution: Won't Fix
I don't think we can really support truly heterogeneous capacity.
the problem is that each node's range is replicated to other nodes, so if you have a single node w/ 2x more capacity than other nodes, you can't just give it 2x the range or its replicas will be swamped.
Moving to a dynamo-like system of multiple virtual nodes per machine might allow this, but IMO the complexity is not worth it.
For long-lived clusters, I would recommend starting w/ relatively high end machines, then keeping capacity (relatively) constant as cost decreases. So heterogeneous in cost rather than capacity.
> Pluggable Load and contrib example
> ----------------------------------
>
> Key: CASSANDRA-476
> URL: https://issues.apache.org/jira/browse/CASSANDRA-476
> Project: Cassandra
> Issue Type: New Feature
> Components: Contrib, Core
> Affects Versions: 0.5
> Reporter: Michael Greene
> Priority: Minor
> Fix For: 0.9
>
>
> With CASSANDRA-385 intellibootstrap, a site-specific definition of Load becomes more useful, especially for those with heterogenous capacity in their cluster.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.