[
https://issues.apache.org/jira/browse/KUDU-1848?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Todd Lipcon updated KUDU-1848:
------------------------------
Component/s: perf
> Simplified dictionary-compression for in-memory data
> ----------------------------------------------------
>
> Key: KUDU-1848
> URL: https://issues.apache.org/jira/browse/KUDU-1848
> Project: Kudu
> Issue Type: Improvement
> Components: perf, tablet
> Affects Versions: Backlog
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
>
> In use cases that insert a lot of dictionary-codable data, we find that the
> MRS fills up rather rapidly, partially due to many repeated copies of the
> same values. It would be relatively simple to keep an in-memory dictionary of
> the most recent N binary values, and on insertion, check if we have recently
> inserted the same value. If so, we don't need to make a second copy of it in
> the MRS arena.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
