[
https://issues.apache.org/jira/browse/HBASE-2468?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12870241#action_12870241
]
Jonathan Gray commented on HBASE-2468:
--------------------------------------
IMO this jira is about adding functionality not necessarily changing the
default functionality. I think non-warming, non-prefetching clients should
still be possible, if not default. Seems odd to make a big change that all of
a sudden doing your existing new HTable(name) could possibly result in the scan
of potentially thousands of rows or more and now you need concept of preventing
this from happening.
If a split, scanning forward one row gives you 50% chance of also finding the
second half of split, don't see what going further than that would give you.
This would be at the cost of single-row optimizations that are possible. Going
forward one extra row sounds reasonable to me but again it's not with impact so
should be optional.
> Improvements to prewarm META cache on clients
> ---------------------------------------------
>
> Key: HBASE-2468
> URL: https://issues.apache.org/jira/browse/HBASE-2468
> Project: Hadoop HBase
> Issue Type: Improvement
> Components: client
> Reporter: Todd Lipcon
> Assignee: Mingjie Lai
> Fix For: 0.21.0
>
> Attachments: HBASE-2468-trunk.patch
>
>
> A couple different use cases cause storms of reads to META during startup.
> For example, a large MR job will cause each map task to hit meta since it
> starts with an empty cache.
> A couple possible improvements have been proposed:
> - MR jobs could ship a copy of META for the table in the DistributedCache
> - Clients could prewarm cache by doing a large scan of all the meta for the
> table instead of random reads for each miss
> - Each miss could fetch ahead some number of rows in META
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
