[
https://issues.apache.org/jira/browse/CASSANDRA-1034?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13088280#comment-13088280
]
Mck SembWever edited comment on CASSANDRA-1034 at 8/20/11 10:10 PM:
--------------------------------------------------------------------
What's the status on this? This issue and its relations back to CASSANDRA-2878
are the only reason we're using OPP. I suspect other users setup with both
cassandra and hadoop (or brisk) could be in the same boat. Not only does OPP
leave an unbalanced ring (i've had a case where all data went to one node
because the keys/tokens were longer than normal) it leaves poor performance to
hadoop jobs as tasks requirement on data locality has become stricter (w/
CASSANDRA-2388). Apart from the plain preference to be using secondary indexes
over OPP.
was (Author: michaelsembwever):
What's the status on this? This issue and its relations back to
CASSANDRA-2878 are the only reason we're using OPP. I suspect other users setup
with both cassandra and hadoop (or brisk) could be in the same boat. Not only
does OPP leave an unbalanced ring (i've had a case where all data went to one
node because the keys/tokens were longer than normal) it leaves poor
performance to hadoop jobs as tasks requirement on data locality has become
stricter (w/ CASSANDRA-2388).
> Remove assumption that Key to Token is one-to-one
> -------------------------------------------------
>
> Key: CASSANDRA-1034
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1034
> Project: Cassandra
> Issue Type: Bug
> Reporter: Stu Hood
> Assignee: Sylvain Lebresne
> Priority: Minor
> Fix For: 1.1
>
> Attachments:
> 0001-Make-range-accept-both-Token-and-DecoratedKey.patch,
> 0002-LengthPartitioner.patch, 1034-1-Generify-AbstractBounds-v3.patch,
> 1034-2-Remove-assumption-that-token-and-keys-are-one-to-one-v3.patch,
> 1034_v1.txt
>
>
> get_range_slices assumes that Tokens do not collide and converts a KeyRange
> to an AbstractBounds. For RandomPartitioner, this assumption isn't safe, and
> would lead to a very weird heisenberg.
> Converting AbstractBounds to use a DecoratedKey would solve this, because the
> byte[] key portion of the DecoratedKey can act as a tiebreaker.
> Alternatively, we could make DecoratedKey extend Token, and then use
> DecoratedKeys in places where collisions are unacceptable.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
