It's the Server, Stupid!

Given that Microsoft formally announced Windows 2000 (Win2K) recently, I should devote this space to Windows 2000 Server Terminal Services. However, if you’re reading this, you’re probably familiar with the concept: Terminal Services is now part of the core OS, RDP now supports client-side printing and a shared clipboard, and the licensing structure differs only slightly from that of Windows NT Server 4.0, Terminal Server Edition (TSE). If you’re not familiar with the concept, you’ll get more from the articles I’ve written for Windows 2000 Magazine or the chapter I contributed to Mastering Windows 2000 Server (Sybex, 2000) than you could from one editorial anyway. So instead, let’s talk about the server-based computing market and how it’s perceived.

Come to think of it, this discussion is about Win2K to some degree. As many have observed, Microsoft’s decision to make multiuser computing a core service of its network operating system (NOS) should boost terminal services use by reducing the risks of adopting the technology and by providing remote administration capabilities, which can benefit even those who are committed to the PC-centric model. But once you get past the tinkering stage and are thinking seriously about whether to deploy Terminal Services on even a small scale, you’ll probably have to convince a higher-up that it’s a good idea.

Trouble is, the higher-up might know the server-based computing model as thin-client computing, which is how the trade press (and this newsletter) often refers to it. (When we started this newsletter, we weren’t quite sure what to call it. Server-Based Computing UPDATE is unwieldy, and nothing else really fit, so we finally decided that people would at least know what we meant if we called it Thin-Client UPDATE.) The trouble with calling the model thin-client computing is that the name puts more emphasis on the client than the server, which is really the opposite of how this computing model looks to the people running it.

Confusing people is rarely a good idea, and it’s worse yet when you’re trying to promote an unfamiliar concept. For enthusiasts who don’t really know the model, thin-client computing implies talking about how well terminals work and how great terminals are. For the less enthusiastic who don’t really know the model, such a name provides an easy way of pointing out that desktop-based applications are far from dead—after all, PCs still outnumber Windows-based terminals (WBTs) on the desktop by more than 2:1, and PC manufacturers aren’t seeing demand for WBTs.

Ironically, I have to agree with the pessimists’ surface argument. Server-based computing isn’t always about scrapping the PC. Sometimes, to be sure, a terminal is a better choice than a PC. However, so long as mainframes are still in use—and probably punch cards, somewhere—no one will get me to say that one computing model will completely edge out another. However, I also think that focusing on the relatively small market for terminals (compared with PCs) misses the point. That most PC OEMs (with the notable exception of Compaq) are leaving the terminal-making to others proves nothing about the viability of the server-based model. Of course PC OEMs aren’t making WBTs—the market is crowded and the unit replacement is low. If I were trying to exploit a market and had a reputation for building PCs and servers, I’d stick with servers, which bear the brunt of computing and are therefore more likely candidates for replacement or expansion than a terminals that are meant to be "set and forget." So yes, Hewlett-Packard has discontinued its Windows terminal line and Dell doesn’t sell a terminal, but both companies build servers for the server-based computing market.

My point is that if you’re trying to introduce terminal services to your organization, be careful not to get too hung up on the terminal side of the equation. There are definite benefits to WBTs—no doubt about it—and they can outperform some PCs for some applications. However, terminals aren’t the right answer for everybody. Server-based computing is about centralized administration, easy interoperability, and uncomplicated deployment of new OSs and applications without upgrading desktop hardware—not necessarily about how fabulous WBTs are. I suspect that you’ll have a lot more luck persuading strong supporters of the PC model that server-based computing can work if they don’t think that you’re trying to take their PCs away. (If you are, you can work on that after you’ve sold them server-based computing.)