Use Thin Clients

A lot of the advice given above applies to thick clients and servers but seems to ignore the huge benefits of thin clients: lower power consumption, footprint, noise and MAINTENANCE. Fanless thin clients will run until the screen resolution has evolved past their limit. They last several times longer than a thick client and avoid the need to walk around. They also cost less to buy because there is less material in them.

Once you are using thin clients you can concentrate on the servers and do whatever you want to keep them updated. Because one server can operate many thin clients this is a much lighter task. The entire system has much better performance because a server can specialize in some application and be tuned for it. You can cache all the files in RAM, for instance. You can use huge RAID or SSD to boost performance, too, at much less cost than optimizing the many clients.

Use thin clients. The savings are huge. I like GNU/Linux on thin clients and terminal servers to save on per-seat charges for licences.

@Robert Pogson

Hey Robert,

Some Interesting thoughts there. I'd like to say that I can't disagree with you, in fact of the 6 "classes" of desktop hardware we deploy at my place of primary employment, the bottom 4 are "Linux thin clients." In truth, they are systems that once ran Windows "on the front line," but have been retired to less arduous services as thin clients running Fedora Core 10. (Real thin clients offer more power savings than rebuilding old PCs, it should be noted. This matters a lot on large deployments!)

As you have stated, one of the primary drivers on this is the desire to save on CALs. In our case, due to an odd quirk of a piece of line-of-business software, we were forced to use VDI for 80%+ of our users. Going the thin client route has saved us a bundle on CAL costs, and on maintenance.

The maintenance savings come in two forms: as thin clients, the systems are truly disposable. There are numerous spares stacked away in case someone’s desktop eats its own head. If a desktop dies, the user can just grab one and swap out their existing box. Secondly, the users don’t comprehend Linux at all, and thus their flailing around trying to beat their thin client into a different shape has no effect. (Using Firefox on the thin client isn’t going to get you to facebook or viral website #422 at work, sorry.)

In fact, we have really pushed it by stretching our desktop estate long past what I would ever recommend because of the use of Thin Clients. Our “lowest class” of deployed desktops is called “disharmonious junk.” It is literally every junker PC that just doesn’t fit into one of the other 5 classes of deployed desktop. While I would never, ever support desktops like that in a fat client scenario, (a few months of support time would pay for the replacement of the 30 or so desktops in this category around here.) As thin clients though…who cares? Once they die, they die, (we don’t repair disharmonious junk,) and a spare is swapped in.

So I guess a long ramble concluded: you are right, thin clients can drive down the cost of maintaining desktops, and indeed it’s something we have embraced. I have found however, that keeping your desktops as homogenised as possible, and thus having identical spares waiting in the wings is a benefit even in a thin-client (or partly thin-client) desktop deployment scenario.