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

Re: Speed issues with ra_neon to remote repositories

From: John Beranek <john_at_redux.org.uk>
Date: Thu, 25 Jun 2009 09:47:42 +0100

Bolstridge, Andrew wrote:
>> -----Original Message-----
>> From: John Beranek [mailto:john_at_redux.org.uk]
>> Sent: Wednesday, June 24, 2009 3:45 PM
>> To: users_at_subversion.tigris.org
>> Subject: Speed issues with ra_neon to remote repositories
>>
> [snip]
>> The directory tree contained 3739 files totalling 29MiB.
>>
>> An import from the UK to the UK master takes: 38s An import from
> France to the
>> France slave takes: 3m48s An import from France to the UK master
> takes: 3m46s
>> An import from India to the UK master: 27m57s !!
>>
> [snip]
>> I retried the test using ra_svn and the India import took just _22s_!
>>
>> OK, so on the surface ra_svn is clearly the solution...except you
> can't do
>> transparent proxying of reads/writes to a local slave with ra_svn...
>
>
> I suppose if ra_svn takes 22s over the WAN, and ra_neon takes 38s on the
> LAN... you might as well use ra_svn all the time on a single UK-based
> repository. Obviously you don't get the other features of using
> Apache-based repositories, but if you can live with that, then you're
> good to go.

Well, indeed SVN access is quicker for commits, wherever you are.
However, if you were to do without mirrors, checkout times from remote
sites will be impacted for all users. For the directory tree used in my
import tests I got the following timings:

HTTP checkout in India from UK master: 17s
SVN checkout in India from UK master: 20s
HTTP checkout in UK from UK master: 6s
HTTP checkout in UK from UK master: 7s
SVN checkout in UK from UK master: 5s

It's interesting here to see that HTTP is quicker than SVN from India -
this might be because the HTTP server is configured with mod_deflate.

> Can you serve repositories using svnserve and apache at the same time?

That was certainly my reading of the situation - authentication
configuration is certainly complicated though. Authentication to a
Windows Active Directory is very easy with Apache, but nowhere near easy
with svnserve+sasl...

> If so, do so, and tell people that large imports will be faster if they
> change the protocol in the url. Most of the time (as you say people
> don't do this very often) they'll happily import the files and go off to
> lunch, or kick it off at hometime. I do this when retrieving files from
> a remote VSS DB - it takes all night, but I don't care as long as it's
> finished by the time I arrive next morning. Most people are used to some
> tasks taking ages, especially network-related ones. We just work around
> it.

Well, I can certainly see this viewpoint, but one problem we have is
that moving to Subversion will generally see an improvement in
speed...except when it doesn't.. Hard to get people to buy into a move
that may slow down the things they find are already slow (with CVS and
per check-in tagging).

> BTW. TortoiseSVN has a new repository graph, I don't know if that helps.
> http://tortoisesvn.tigris.org/tsvn_1.6_releasenotes.html#revisiongraph

I'll have to take a look at the latest version - but a single platform
graph viewing tool really doesn't quite cut it. We have quite a few
Linux users - also, a lot of users currently use the web-based cvsgraph
functionality in cvsweb, and I've seen nothing like this for Subversion.

John.

-- 
John Beranek                         To generalise is to be an idiot.
http://redux.org.uk/                                 -- William Blake
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=2365207
To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_subversion.tigris.org].

Received on 2009-06-25 10:49:01 CEST

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