On Nov 24, 2003, at 1:47 PM, Ryan Hunt wrote:
> It seems to me that a repository holding a large number of large files
> being accessed by a large number of people will present a bottle neck
> of accessibility since it appears as though there is a limit of about
> 30-50MB/min of transfer out of the repository regardless the number of
> clients. They all seem to share a slice of that 30-50MB pie.
>
> So my question is, can a repository be distributed over several
> systems to increase processing power? A cluster perhaps? I know that
> NFS mounted filesystems can cause repository corruption so am not sure
> if clustering would as well.
Clustering the present implementation is right out; it's just not
designed to sync properly. Making that work is something we'd like to
do in the future, but there's no schedule for that work at this time.
You might look into svk. It's not quite what you want either, but
perhaps closer than svn:externals is. svk manages a galaxy of
loosely-coupled SVN repositories. Each of them would still be subject
to the bandwidth limitations you raise here, but you could put
development on different branches on different servers well enough. It
might cost just as much (or more) to sync the repos as just to work in
them directly, though: you have to control that by choosing what and
when to sync (this is not 100% all the time replication like, say,
ClearCase MultiSite).
http://svk.elixus.org/index.cgi?HomePage
-==-
Jack Repenning
CollabNet, Inc.
8000 Marina Boulevard, Suite 600
Brisbane, California 94005
o: 650.228.2562
c: 408.835-8090
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Wed Nov 26 04:26:48 2003