[
https://issues.apache.org/jira/browse/HIVE-12236?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14983444#comment-14983444
]
Gunther Hagleitner commented on HIVE-12236:
-------------------------------------------
This doesn't seem to be related to LLAP. This is just an addition to
SimpleFetchOptimizer.
I take it you called out "in_file" because it's pretty likely that doing that
in parallel wins over doing it in HS2? What about others though? Some string
manipulations can be slow. Locate, levenstein distance, a bunch of those
combined can just be as bad (and create lots of new objects).
The metadata queries don't use any builtins anyways, do they?
> LLAP: Prevent metadata queries from thrashing LLAP cache
> --------------------------------------------------------
>
> Key: HIVE-12236
> URL: https://issues.apache.org/jira/browse/HIVE-12236
> Project: Hive
> Issue Type: Improvement
> Components: Physical Optimizer
> Affects Versions: 1.3.0, 2.0.0
> Reporter: Gopal V
> Assignee: Gopal V
> Fix For: 1.3.0, 2.0.0
>
> Attachments: HIVE-12236.WIP.patch
>
>
> Currently, metadata queries fired by BI tools tend to thrash LLAP's cache.
> Bypass the cache and process metadata queries directly from HiveServer2
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
