SVN, large files & performance

Hi,

First of all I want to thank you.

No no no. I HAVE to thank you for your astonishing work that has supported
so many productions without failure. As a matter of fact, the architectures
that I designed around SVN have only suffered human mistakes so far.

Today, I'm doing some performance tests with svn and large binary files.

I have to commit large files with a content that changes a lot but the whole
process is very slow.

It seems to me that it is mainly due to the binary diff operated by the
client with its local base copy to only send differences.

Am I right about this? Could I get performances close to a samba copy once
disabled?

I could deal with the whole file being stored instead of the diff. What
matters to me is performance.

Currently, I use TortoiseSVN as a client for my tests but will need to write
my own client application to interpret custom properties. Programmatic are
more than welcome.

Could you point me out the best way to get around this problem?

All the best.

Michel Courtine | Lead Tools Developer | MKO Games
La Maison | 13/15 rue Gaston Latouche | 92 210 | St-Cloud | France
Tel: +33 1 78 16 5861 | Fax: +33 1 41 12 2001
 <http://michak.net/> http://michak.net

 
Received on Mon Oct 2 06:14:07 2006