[
https://issues.apache.org/jira/browse/PHOENIX-3796?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15975570#comment-15975570
]
James Taylor commented on PHOENIX-3796:
---------------------------------------
[~rajeshbabu] - good point about the visibility of the table rows prior to the
index table being updated. There's already the inverse of this issue, though -
the MVCC lock on the index table regions would be released before the MVCC lock
on the data table, so the index data would start to be visible before the data
table. It's somewhat impossible not to have these small windows given that with
global indexes there are two completely separate hbase tables. I think with
local indexes, we can make some good guarantees with 1.3 and above. We should
document this behavior with global indexes.
> LocalIndexes apply the entire batch for each mutation in a batch
> ----------------------------------------------------------------
>
> Key: PHOENIX-3796
> URL: https://issues.apache.org/jira/browse/PHOENIX-3796
> Project: Phoenix
> Issue Type: Bug
> Reporter: Lars Hofhansl
> Assignee: Lars Hofhansl
> Attachments: 3796-4x.txt, PHOENIX-3796-4.x-HBase-0.98.patch,
> PHOENIX-3796.patch
>
>
> [~mujtabachohan] pointed out an "exponential" increase in update time in the
> size of the batch.
> Indeed when I look at the code in Indexer.java I see that for each call to
> postPut or postDelete we do insert the entire into the local index. (So for a
> batch of size 10000 we'd make 10000 index updates 10000 times)
> [~rajeshbabu]
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
