Distributed Development Locations

[Kevin Stark <ke...@hotmail.com>]

I have a question about the use of Subversion in a distributed development environment.  We're considering making a move to Subversion as our SCM tool, however we have several development locations throughout the world (India, Brazil, Ireland, USA, Argentina) all of which can work on any one product simultaneously. 

Obviously, one of the good things about Subversion is its ability to handle a ton of check-in/check-out activity from just about any location that can speak HTTP. That being said, I'm wondering about how well Subversion will work with a centralized server configuration when the development locations are highly distributed and use less than optimal network links. Are there possible configurations where there is a multiple-master server configuration, a la Active Directory, meaning that there are multiple servers that host the code and changes are replicated between them? Has anyone had success deploying in highly distributed commercial environments like this? We would have approximately 25K files, 5M LOC, and 20 GB of total data being managed.

Please include my e-mail address in your replies as I am not a member of the mailing list yet.

Kind Regards,

Kevin Stark

Re: Distributed Development Locations

[William Nagel <bi...@stagelogic.com>]

> I have a question about the use of Subversion in a distributed 
> development environment.  We're considering making a move to 
> Subversion as our SCM tool, however we have several development 
> locations throughout the world (India, Brazil, Ireland, USA, 
> Argentina) all of which can work on any one product simultaneously.
>   
> Obviously, one of the good things about Subversion is its ability to 
> handle a ton of check-in/check-out activity from just about any 
> location that can speak HTTP. That being said, I'm wondering about how 
> well Subversion will work with a centralized server configuration when 
> the development locations are highly distributed and use less than 
> optimal network links. Are there possible configurations where there 
> is a multiple-master server configuration, a la Active 
> Directory, meaning that there are multiple servers that host the code 
> and changes are replicated between them? Has anyone had success 
> deploying in highly distributed commercial environments like this? We 
> would have approximately 25K files, 5M LOC, and 20 GB of total data 
> being managed.

Since I don't know what you're moving from, I can't speak to whether 
Subversion will perform better with widely distributed development 
locations.  That said, Subversion is designed to perform very well in a 
setup with a central server and widely distributed clients.

* It keeps local pristine copies of all files in a working copy, so 
many comparisons and status/metadata checks can be performed without a 
round trip to the server.

* All commits and updates are transfered as diffs against the local 
version.  This applies to binary files as well as text files.


If this isn't sufficient, you might try taking a look at SVK 
(http://svk.elixus.org/).  It's a decentralized version control system 
that is built on top of Subversion.  It's still in beta, but it does 
seem to be moving along nicely.  (DISCLAIMER: I haven't actually used 
it, but I've heard good things)

-Bill

>  
> Please include my e-mail address in your replies as I am not a member 
> of the mailing list yet.
>  
> Kind Regards,
>  
> Kevin Stark
>  
>  


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