[
https://issues.apache.org/jira/browse/HIVE-9729?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14344404#comment-14344404
]
Sergey Shelukhin commented on HIVE-9729:
----------------------------------------
The approach from the description will be taken
> LLAP: design and implement proper metadata cache
> ------------------------------------------------
>
> Key: HIVE-9729
> URL: https://issues.apache.org/jira/browse/HIVE-9729
> Project: Hive
> Issue Type: Sub-task
> Reporter: Sergey Shelukhin
> Assignee: Sergey Shelukhin
> Fix For: llap
>
>
> Simple approach: add external priorities to data cache, read metadata parts
> of orc file into it. Advantage: simple; consistent management (no need to
> coordinate sizes and eviction between data and metadata caches, etc);
> disadvantage - have to decode every time.
> Maybe add decoded metadata cache on top - fixed size, small and
> opportunistic? Or some other approach.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
