You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@couchdb.apache.org by Jens Alfke <je...@couchbase.com> on 2012/07/01 21:10:18 UTC
HTTP Pipelining?
Does the HTTP server in CouchDB [MochiWeb?] support HTTP pipelining? I'd like to enable it in TouchDB's replicator as it should make 'pull' replication a lot faster, but I first want to make sure it's not something unsupported or flaky in CouchDB.
In the long run it would be good to come up with an extension to the API to allow the replicator to fetch remote revisions in bulk. Currently the large number of small HTTP GETs (one per revision) greatly lower the performance of pulls.
—Jens
Re: HTTP Pipelining?
Posted by Paul Davis <pa...@gmail.com>.
Pretty sure Bob's got it hit on the head here. Mochiweb should handle
it just fine but its quite possible that we have latent bugs since I'm
not sure how many people have exercised this bit of the code.
On Sun, Jul 1, 2012 at 5:42 PM, Robert Newson <rn...@apache.org> wrote:
> It should, yes. Mochi *does* but there have been some spots in our handling where we misbehaved. If there *are* bugs, I would really like to squish them.
>
> On 1 Jul 2012, at 15:10, Jens Alfke wrote:
>
>> Does the HTTP server in CouchDB [MochiWeb?] support HTTP pipelining? I'd like to enable it in TouchDB's replicator as it should make 'pull' replication a lot faster, but I first want to make sure it's not something unsupported or flaky in CouchDB.
>>
>> In the long run it would be good to come up with an extension to the API to allow the replicator to fetch remote revisions in bulk. Currently the large number of small HTTP GETs (one per revision) greatly lower the performance of pulls.
>>
>> —Jens
>
Re: HTTP Pipelining?
Posted by Robert Newson <rn...@apache.org>.
It should, yes. Mochi *does* but there have been some spots in our handling where we misbehaved. If there *are* bugs, I would really like to squish them.
On 1 Jul 2012, at 15:10, Jens Alfke wrote:
> Does the HTTP server in CouchDB [MochiWeb?] support HTTP pipelining? I'd like to enable it in TouchDB's replicator as it should make 'pull' replication a lot faster, but I first want to make sure it's not something unsupported or flaky in CouchDB.
>
> In the long run it would be good to come up with an extension to the API to allow the replicator to fetch remote revisions in bulk. Currently the large number of small HTTP GETs (one per revision) greatly lower the performance of pulls.
>
> —Jens