You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@subversion.apache.org by Derek Beatty <De...@Sun.COM> on 2006/04/13 21:10:35 UTC

max number of developers on one repository

What is the maximum number of developers who can safely use a single  
repository?

We're looking at perhaps 3-4 active release branches, a handful of  
feature branches, and a couple dozen gig of data.

Although server size will be a question, it is not an issue for us.

I'm more interested in real-world experience than theoretical arguments.

Thanks,


  Derek Beatty                                   +1 512 401 1221
  Processor CAD, Sun Microsystems, Inc.    fax:  +1 512 401 1490
  Austin, Texas                           cell:  +1 512 297 6235

NOTICE: Unencrypted email is not secure and may not be authentic. If  
you have any doubts about this message please phone me to confirm.



---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

Re: max number of developers on one repository

Posted by Konrad Rosenbaum <ko...@silmor.de>.
On Thursday 13 April 2006 23:10, Derek Beatty wrote:
> What is the maximum number of developers who can safely use a single
> repository?

indefinite. There are repositories out there with several hundred registered 
developers, eg. KDE has about 800 (maybe 900 by now).

> We're looking at perhaps 3-4 active release branches, a handful of
> feature branches, and a couple dozen gig of data.

From reading the list: make sure this data is in a sane directory structure 
- svn seems to have problems with directories that contain thousands of 
files (up to about a hundred is not a problem).

> Although server size will be a question, it is not an issue for us.

Depends on how many simultaneous accesses you expect to have. If you access 
it from the Internet (or VPN) you usually exhaust the bandwidth of the 
outside line faster than the CPU-power of the server.



	Konrad