VersionControl for huge codebase

Our project's codebase is more than 9GB. Mostly Cobol, Pro*Cobol & Java files along with other config files. We are currently using SVN to manage it and during integrated check-outs and builds SVN's performance is bad. For example to checkout the complete sources it takes >4hrs and if we commit say 12 or more files it takes close to 开发者_高级运维30min. Appreciate suggestions on how to tune SVN or any alternate open source version control to handle this volume of codebase. Thanks

-RamVenkat

It may well be your network or server is not as good as you think. I have a 300,000 revision repo that is 12 Gb (last time I looked)(I don't actually know how much that is when checked out locally!) running on an underpowered VM. I wouldn't expect any SCM, distruibuted or centralised, to checkout a fresh copy of all that in a couple of minutes.

4 hours on the other hand is just plain broken, and 30 min checkins - something else is bad for you. You need to find that first or migrating to git will still be broken. Look at cpu and memory usage on the server, look at network performance.

SVN does offer you a few features to help with your problems though, take a look at sparse directories which allow you to checkout a partial copy of the repo and extend your WC as needed. You don't need to checkout all the stuff you never need.

Short answer, actually just one word: git. Or to be not so straight forward: "Why don't you try and evaluate git?" see: http://git-scm.com/

There are tools to migrate from svn to git which should make it easy for you to start testing. look at: http://www.jonmaddox.com/2008/03/05/cleanly-migrate-your-subversion-repository-to-a-git-repository/