[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: Greg Stein <gstein_at_gmail.com>
Date: Mon, 10 Nov 2008 08:16:29 -0800

On Mon, Nov 10, 2008 at 8:02 AM, Mark Phippard <markphip_at_gmail.com> wrote:
>...
> Would one scale better on the server? There are a lot of Subversion

The two should be about the same difficulty for the server. In one
case, the HTTP request is your delimiter, and in the glom, you make up
something else. (as Julian Reschke noted)

> hosting services out there. If commits are performing multiple
> requests on the server does that create a scalability problem? The
> amount of work/data seems pretty similar, but I am wondering about the
> increase in number of connections as well as any extra work the server
> has to do to handle the requests and manage the transaction.
>
> This is more a question than a statement. Would the scalability of
> the server likely be impacted by this decision, and if so, to what
> degree?

I would posit that multiple requests can be scaled better.
Specifically, with Google's backend, I could see how those PUTs could
be directed to multiple servers so they could be handled almost in
parallel. (there are several reasons in Google's design that isn't
preferable, but I can *envision* the scenario where it would be
Hotness)

For us lame mortals without that infrastructure, nah... there aren't
really any factors in the decision which would affect scalability.
(that I can think of)

Cheers,
-g

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-11-10 17:16:48 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.