[
https://issues.apache.org/jira/browse/HADOOP-9151?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13536806#comment-13536806
]
Binglin Chang commented on HADOOP-9151:
---------------------------------------
Yes, it makes sense to avoid buffer copy. But I suspect the performance penalty
of buffer copy isn't that much(rpc call should be small in most cases), but the
penalty of complex code logic may be more.
Plus that the overhead of creating new proto objects for every rpc may already
take a lot more resource than buffer copy, so buffer copy may be not the
bottleneck at all.
And another concern is this protocol may be bad for async non-blocking io,
which just want a total length, receive enough info to process, then call rpc
handles. More distinct packats make more handlers/callbacks.
> Include RPC error info in RpcResponseHeader instead of sending it separately
> ----------------------------------------------------------------------------
>
> Key: HADOOP-9151
> URL: https://issues.apache.org/jira/browse/HADOOP-9151
> Project: Hadoop Common
> Issue Type: Sub-task
> Reporter: Sanjay Radia
> Assignee: Sanjay Radia
> Attachments: HADOOP-9151.patch
>
>
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
