Despite all the network computer hype created by Oracle Corp., Sun Microsystems and others, the market for intelligent desktop devices that load and process server-based applications hasn't really amounted to much. But that doesn't mean the appeal of thin-client computing isn't growing. In fact, a sizable market for Windows terminals has emerged during the past two years. Most of the credit for nurturing this growth goes to Citrix Systems and its incredibly successful WinFrame, the first product to deliver multiuser terminal capabilities to Windows NT.

To view the Report card.
While Citrix was enjoying its initial success with WinFrame running under NT 3.5, expanding and enhancing its ICA (Independent Computing Architecture) thin-client display protocol, Microsoft Corp. couldn't quite make up its mind whether Citrix was a friend or a foe. On the one hand, Microsoft trumpeted WinFrame as yet another reason for corporations to embrace NT. On the other, it denied Citrix a source-code license for NT 4 and flirted for a while with the prospect of developing its own version of multiuser NT. Eventually, the two companies joined forces, leading to the release of Microsoft's Windows NT Terminal Server Edition (previously known as Hydra) and Citrix's MetaFrame, which provides enhancements to NT Terminal Server Edition, including support for Citrix's ICA protocol. Unfortunately, for organizations interested in deploying Windows terminals, the Citrix-Microsoft relationship is a source of considerable frustration: Neither vendor alone offers a complete solution (see "Microsoft Windows NT Terminal Server or Citrix WinFrame/MetaFrame?" at www.networkcomputing.com/ 920/920r1.html).

Probably this year's biggest news in the Windows terminal market has been the adoption of Windows CE as the OS of choice for Windows-based terminals (WBTs). In one of Network Computing's Real-World Labs® at Syracuse University, we tested five CE-based WBTs in various stages of development. Despite a number of technical problems--which we expected in light of the immaturity of the underlying CE 2.1 system software--we have little doubt that these devices eventually will achieve widespread acceptance among small and large organizations alike. Because they are Windows-centric, the products can meet the application requirements of the majority of today's knowledge workers. In addition, they dramatically reduce the administrative costs associated with today's Windows-based network environments.

Unfortunately, despite the obvious appeal of a standard, small footprint, multitasking, ROM-loadable OS serving as the foundation for WBTs, it is too early to embrace these offerings for large-scale deployment. Performance falls far short of that found in earlier-generation, ICA-based devices, which are based on simpler and more efficient operating systems. While reluctant to go on record with specifics for understandable reasons, vendors expressed concern about bugs and functional limitations in CE, which, in spite of its official designation, is clearly a .0 release.

For example, the current version of CE does not include a mechanism for securing the local client configuration from changes by the user--a capability that would appear to be fundamental to a thin-client rollout. Both Neoware Systems and Boundless Technologies cite this as a requirement and are committed to adding this capability on their own before they will ship a product. And though all the vendors have ambitious plans for central configuration management and automated updates of client applications, the management tools needed to support large-scale deployment are not yet in place.