[
https://issues.apache.org/jira/browse/COUCHDB-1743?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13644567#comment-13644567
]
Robert Newson commented on COUCHDB-1743:
----------------------------------------
Are these ideas backed with benchmark data?
Moving to more stateful view servers sounds wrong to me because couchjs
processes are used for many different view groups over time.
Where are the bottlenecks in the current view server protocol? I think we all
agree that we can make this faster but I'd like to see the discovery phase
first.
Finally, speeding a single view index is a laudable goal but let's not lose
sight of the fact that with lots of view groups building in parallel we do
exercise multiple cores quite well. If each of those view servers is internally
parallel, we'll impact that.
Paul Davis has suggested a single couchjs process per couchdb node (rather than
a pool of them) that internally manages the multiple view group states. That
plus some batching or pipelining of requests gets you a very long way without
great changes to the protocol.
Just my 2c but I'd love to see data.
> Make the view server & protocol faster
> --------------------------------------
>
> Key: COUCHDB-1743
> URL: https://issues.apache.org/jira/browse/COUCHDB-1743
> Project: CouchDB
> Issue Type: Improvement
> Reporter: Dave Cottlehuber
> Labels: couchdb, erlang, gsoc2013, html, javascript, nodejs, rest
>
> View server protocol enhancements/refactoring - unix sockets, pipelining,
> different wire format etc. Faster!!
--
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
