[
https://issues.apache.org/jira/browse/PHOENIX-1247?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14129156#comment-14129156
]
Brian Johnson commented on PHOENIX-1247:
----------------------------------------
Yea, and I guess I mixed in some group by stuff as well, but I don't know how
group by is currently implemented. Maybe this is relevant and maybe it's not?
> Join using sorted data
> ----------------------
>
> Key: PHOENIX-1247
> URL: https://issues.apache.org/jira/browse/PHOENIX-1247
> Project: Phoenix
> Issue Type: New Feature
> Reporter: Brian Johnson
>
> Similar to pig merge join, Phoenix should have a join that takes advantage of
> the sorted nature of hbase keys. If you have two tables that have a column
> which is sorted the same as the rowkey, you can join them efficiently without
> keeping either table in RAM. This also depends on using a split policy which
> ensures the keys will be in the same region like
> DelimitedKeyPrefixRegionSplitPolicy
> As an example, we keep user data in hbase where the first part of the key is
> the user id and the second part makes it unique for each event. We then have
> a column which is just the user id which will always be sorted because of the
> rowkey
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
