[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: HTTP protocol v2: single commit request, or many PUTs?

From: Philipp Marek <philipp.marek_at_emerion.com>
Date: Wed, 12 Nov 2008 09:09:51 +0100

On Montag, 10. November 2008, Ben Collins-Sussman wrote:
> Let me revise the lists, leaving out the parallel PUTs, since we have
> no plans for that anytime soon.
>
> Multi-request motivations:
>
> * opens the door for pipelined writes, the way ra_svn does it
> * more resilient on a flaky network
I don't really understand you here.

If you have packet loss, then a single TCP connection will stall; but if you
do 3 in parallel, then the 4th has to wait for one of the 3 to finish - and so
you'll be waiting again.

TCP already has the needed guarantees; and I don't see subversion doing
several tries getting a stream into the repository, just because the first
connection failed.

> * more detailed logging
> * don't need to change much code
>
> Single-request motivations:
>
> * one request means less traffic overall
> * can use standard libsvn_repos commit-editor
>
> Hmmm. Maybe multi-request *is* more advantageous. Let's let this
> thread sit around for 24 hours and see if people raise other points.
My gut feeling goes into multiple connections ... but more because I'd hope
that they could be done in parallel to multiple servers, with a single doing
the commit processing when all has been sent.

Regards,

Phil

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-11-12 09:10:15 CET

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.