I have not seen very much impact, but then, I don't readily check that on the clients, I just wait on the clients to complain. I've noticed that our test groups tend to report a much lower impact than what we actually see, I think this is because of several things. Mostly, and I don't know how you tested it, the tests here tend to be done in a perfect scenario, with little to nothing going on, on the client but a backup, nothing going on on the network but the backup, nothing going on, on the master but the backup, ideal conditions. I think when you get to a real world scenario with applications sharing that same resource set that the backups use on the client it tends to take more to do the same things. But that is just what I've seen.