My company just opened a satalite office in another state. I am trying to make it as easy as possible to keep everything between the two offices a close as possible. Anyone do this across OS's? The main file server here is Linux Enterprise and the server offsite is Windows 2003. Did you use RSYNC? The main issue I see is that "what if" situation where people access the same file on different servers and they both go to update. How have you handled this?

I might try to do it this way: Setup a share specifically for projects working on that remote site on the site's local Win2K3 box and then point that server to a Samba DFS root housed on the Linux box for the other data. Replicate the satellite office's data to the DFS root if any people at the home site need its data, this should keep unnecessary traffic to a minimum. Check out Samba with DFS support...I think DFS will be your best bet. I don't think there's ever going to be a way to be 100% certain that no two people are trying to work on the same file in two different offices although if you only have one data set you'll be in pretty good shape. This is why I say have the remote site point to a DFS root that stores the master replica and work with the data that way. You can setup certain shares to replicate their entirety to the server in the satellite office...such as an apps share where people won't be changing data, maybe have it replicate once a week on off hours or just do it manually when you need it same for backups. I believe DFS can track changes and replicate only changes to data incrementally so that a DFS share can be replicated from master and track changes to make sure the two shares are always in sync.

Ok sorry I should have given more info.

It started out that these offices would be working one there own jobs so the only reason to mirror would be to back up the other server at night. Fast forward to the vision now is that both servers will work and update regularly in both directions so that both offices will be able to look at hypothetically the "same" files.

DFS would be a good bet...my post above still applies.