[
https://issues.apache.org/jira/browse/CASSANDRA-749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12855387#action_12855387
]
Jonathan Ellis commented on CASSANDRA-749:
------------------------------------------
Took a look at the HBase stuff. Like Stu said, they have both local indexes
(IHBase) and distributed ones (THBase). Local indexes were implemented second,
but seem to be regarded as better for most situations since they are
substantially faster. (Neither THBase nor IHBase does materialization, so that
is an important difference from what we are proposing here.)
IHBase indexes are built in memory on regionserver start, and are not persisted.
THBase stands for transactional. 2PC is used to keep indexes in sync with the
original data to avoid inconsistency problems, but "you cannot rely on the
transactional properties in the face of node failure." A lot of the tickets
dealing with THBase must be searched for under the term OCC [optimistic
concurrency control], including the first one, HBASE-669.
IHBase and THBase are now both part of the same "transactional" contrib
package. I'm not sure if you can use both types of indexes in the same CF or
Table.
>From my reading, THBase does not deal with the "very large index rows"
>problem. Possibly rows can already be arbitrarily large under HBase + HDFS?
>(Note that for us rows will always be limited to the size of a single
>machine's local disk, even when we fix the "fit in memory' limitation.)
> Secondary indices for column families
> -------------------------------------
>
> Key: CASSANDRA-749
> URL: https://issues.apache.org/jira/browse/CASSANDRA-749
> Project: Cassandra
> Issue Type: New Feature
> Components: Core
> Reporter: Gary Dusbabek
> Assignee: Gary Dusbabek
> Priority: Minor
> Fix For: 0.8
>
> Attachments: 0001-simple-secondary-indices.patch,
> views-discussion-2.txt, views-discussion.txt
>
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
