Once you've got an idea of how much machine you need, what do you specifically buy? For us, the upgrades that made the most sense for designers were dual 1.8 Ghz G5 PowerMacs. Now, why would I go out and buy one of the dual setups, when I've just gotten done saying how they're not worth the money? Simple really, it was the cheapest option available at the time due to a massive shortage on G5 PowerMacs and they weren't producing a single processor G5 when we needed the machines. Even with choosing these machines, it was still a two month wait to get them after we had placed our order. The machines were pretty much stock, except that we had an extra gigabyte of ram put in to them. This is one point that I can't emphasize enough. Always and absolutely make sure that you have budgeted in at least a gig of ram for a design machine. OS X is a massive hog on resources and even the 512 megs that comes stock with these machines can barely power the operating system, let alone the heavy-duty applications your designers will be using. With the extra gig of ram, we bumped up the G5's to 1280 megs of ram. While we only have a short amount of testing on them at this point, we've yet to find that that was not enough ram.
Our eMac machines were also completely stock from the manufacturer, except for the fact we bumped them up to 512 megs of ram. As I said before, OS X is a resource hog and most users are extremely bad about having every application available to them open at the same time. Naturally, you tell them not to do it, but they always end up going down that same street and thus, the need for half a gig of ram is needed and probably more down the road.
You might be asking at this point why I haven't brought up using X Server machines? There are three big reasons for this:
1. Even if we hadn't switched sales over to Windows, we would always have an accounting department running Windows and thus the need for Windows machines. Instead of fighting this and trying to limit the Windows invasion of our offices, I figured we might as go with it because really, all we needed the servers for were file serving and running a couple of server-based applications (Font Reserve Server, QLA, and a backup system.) Mail and Web were already on Linux machines.
2. The cost of the X Server is ridiculous. You can get Windows-based server for a significant cost savings and in all honesty have a machine that does the same exact thing. While a Windows server isn't going to look as "neat" in a server rack, this really doesn't matter much when this thing spends most of its time in the dark room and just needs to do its job.
3. OS X Server is not a finished system. It will do most of the things it needs to do, but at the same time, I get the feeling that Apple really wanted to create something that was their own, which I don't understand, since taking what is great about Windows and Linux would have been a much more worthwhile endeavor. The net result is a system that's about 80% completed. A production environment is no place to reinvent the wheel on a daily basis and I'll choose a Windows machine, because despite all their faults they've worked out a lot of kinks in the 10+ years of creating server software.