[
https://issues.apache.org/jira/browse/HADOOP-1841?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
dhruba borthakur updated HADOOP-1841:
-------------------------------------
Attachment: asyncRPC.patch
Hi Enis, I took the liberty of writing some code to do asynchronous RPC
responses from the server. This patch has the following:
1. There is *one* additional Responder thread, whose job is to write the RPC
responses back to the RPC clients. It does non-blocking IO on the socket.
Because it does non-blocking IO, only one Responder thread is good enough.
2. The RPC handler threads queues the responses of the RPCs into buffers
associated with the response-channel. These responses are flushed to the client
by the Responder thread described above.
3. A test case that creates 50 threads and issues concurrent RPCs. Each of
these RPCs transfers 400K of data from client to server and back from server to
client.
This is very preliminary code but I would like to enhance the test to measure
whether this change is making any difference to performance.
> IPC server should write repsonses asynchronously
> ------------------------------------------------
>
> Key: HADOOP-1841
> URL: https://issues.apache.org/jira/browse/HADOOP-1841
> Project: Hadoop
> Issue Type: Improvement
> Components: ipc
> Reporter: Doug Cutting
> Assignee: Enis Soztutar
> Attachments: asyncRPC.patch
>
>
> Hadoop's IPC Server currently writes responses from request handler threads
> using blocking writes. Performance and scalability might be improved if
> responses were written asynchronously.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
