[
https://issues.apache.org/jira/browse/PHOENIX-4386?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16257940#comment-16257940
]
Lars Hofhansl commented on PHOENIX-4386:
----------------------------------------
Actually with this change I see KeyValueUtil.getEstimatedRowMutationSize again
with 6.9% of the client's CPU time, whereas before the size estimation did not
show up at all (too small to even caught by the sampler).
This is almost as time as it takes to plan each upsert query (9%).
Is this much better than the previous way where we estimated by fractions?
> Calculate the estimatedSize of MutationState using Map<TableRef,
> Map<ImmutableBytesPtr,RowMutationState>> mutations
> -------------------------------------------------------------------------------------------------------------------
>
> Key: PHOENIX-4386
> URL: https://issues.apache.org/jira/browse/PHOENIX-4386
> Project: Phoenix
> Issue Type: Improvement
> Reporter: Thomas D'Silva
> Assignee: Thomas D'Silva
> Fix For: 4.13.1
>
> Attachments: PHOENIX-4386.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
