Print this Page E-mail this URL

For personal productivity applications, Microsoft Office is king and there's little chance of short-term change. A few vendors, including Applix, Corel and Lotus Development, continue to develop thin alternatives to Office. But while these applications are improving in features and functionality, most aren't even on the radar screen of IT managers, let alone their users. It's ironic that a typical user, who takes advantage of no more than 10 percent or 20 percent of the Office feature set, would be so resistant to alternatives, but it is precisely the comfort of that small feature set that keeps users loyal to Microsoft.

While Microsoft is in no danger of losing its dominance in personal productivity to thinner alternatives, Office is relatively easy to deploy via remote-presentation technology. It runs well, even across low-bandwidth connections, and the application processing is sufficiently bursty that a large number of typical users can be supported concurrently by modestly priced servers.

Groupware is a different story, and is clearly moving toward a thinner application environment. Many users have strong ties to Outlook, Notes and other PC-based e-mail, calendar, scheduling and personal information-management applications. However, the ties aren't nearly as strong as the ones to Word and Excel, especially if the functionality of legacy groupware applications is limited or the architecture ties users to a specific desktop computer. Further, all the mainstream groupware vendors and a number of startups are investing heavily in Web- and Java-based implementations. ISPs and Internet portal sites are all pushing the envelope, investing substantial resources in search of Web-based solutions that can lock customers into their service. Go Network, Netscape and Yahoo all offer functional services that can scale to support thousands if not millions of users. These developments, together with stronger orientations toward Internet messaging standards, including IMAP for e-mail, ICAL and VCAL for calendaring and a compelling need by many users for location-independent access, are likely to spell obsolescence for traditional fat-client groupware applications.

Then there are the enterprise applications--ERP and a range of other transaction-processing and decision-support systems, a hodgepodge of legacy host-based and emerging client/server environments. This is prime territory for thin-client deployment and there are big bucks at stake. For legacy host-based applications, both Windows terminals and network computers offer built-in terminal-emulation capability. A potentially huge market exists for Web- and Java-oriented host access for client devices of all sorts. Leading vendors of PC terminal-emulation software, including Hummingbird, Persoft and WRQ, are all actively developing and promoting Web- and Java-based alternatives that allow for centralized distribution of updates and easier management of user profiles.

The ERP vendors, including Oracle, PeopleSoft and SAP, are struggling with many deployment issues; high on the list are fat-client software distribution and performance over slow WAN links. Their responses include both tactical and strategic elements. On the tactical side, they have aligned themselves with Microsoft and Citrix, trying to make their traditional, poorly designed fat clients more accommodating to remote presentation. This is a boon for many organizations that have struggled to deploy and update client application code. At a more strategic level, it is clear that many of the client functions for these application environments will find their way to your browser before long. There's no guarantee that the thin applications will be easier to use than their Windows predecessors, but IT managers will find them much easier to deploy and manage.

Although it now represents less than 10 percent of the total market of thin-client application deployment, there's no question that Windows terminal technology is growing fast, partly because it fills an immediate need and partly because it is maturing. While much attention has been focused on an expanding array of terminal devices, software developments are likely to have more impact on IT pros.

Commitments to Windows CE
Even as prices are falling rapidly and there is competition in the sub-$500 market, Windows terminals have not achieved commodity status. Boundless Technologies, NCD and Wyse Technology dominate the Windows terminal market and each has made significant commitments to Windows CE as an underlying OS. When we tested these devices in November (see "Windows-Based Terminals: Construction in Progress," www.networkcomputing.com/920/920r1.html), we preferred NCD's ThinStar, but all vendors have made improvements to their CE-based product lines since then. NCD has since released the ThinStar 300, which is based on Intel's Lean-Client specification, a less than earth-shattering design that integrates Pentium processors into thin-client computers. Wyse has ratcheted up performance on its 3000-series devices by moving to a very fast Cyrix MediaGX processor, and Boundless has released the Capio, a high-performance device priced at less than $400. Cedar Systems, Esprit and Televideo, among others, have also released new CE-based terminals.

Windows CE has considerable appeal as a standard thin-client OS because it offers a ROM-bootable, 32-bit multitasking kernel with support for a number of CPU alternatives and device-driver support for a variety of hardware. It's not an especially robust OS, but it is clearly good enough for this class of device. Since it will eventually prove difficult to differentiate CE-based offerings in what is likely to be a low-margin commodity market for Windows terminals, the leading vendors are beginning to hedge their bets by introducing new thin-client devices that offer greater flexibility and non-CE architectures. NCD has long offered higher-end client devices and both Wyse and Boundless are working on new terminals built around a derivative of Linux. Neoware Systems also offers both CE and non-CE thin-client products. Ironically, Microsoft may be driving this market for non-CE devices because of relatively high OS software costs and restrictions placed on CE licensees, including prohibitions on vendors from including a local browser or Java Virtual Machine.

And what of the big names? PC stalwarts such as Compaq Computer Corp., Dell Computer Corp. and Gateway have ignored the thin-client market, but IBM says its sales of thin-client devices are doubling every quarter. While Hewlett-Packard Co. is in the Windows terminal market through an OEM arrangement with Wyse, IBM has no plans to deliver a CE-based device. Instead, IBM has built on its initial OEM relationship with NCD and has begun investing more heavily in an internal software and systems architecture for thin clients that is also Unix-based. Given its size and host-oriented history, this is a natural market for IBM to control, and its presence provides further credibility with enterprise customers. Meanwhile, Sun Microsystems, which got the industry real wet with its JavaStation splash, continues to refine a product that is long on potential, but just a bit too revolutionary for today's market.

Pervasively Thin
Just as the invention of Ethernet enabled the PC to become a communications- rather than computation-oriented device, the emergence of wireless networking is likely to do the same for thin-client computing. Increasingly, people will want access to applications and information whenever they like, from wherever they happen to be. Thus, providing access to key applications via wireless networks will become more important. Unfortunately, the wireless metropolitan area networks that dominate today's market will continue to offer very limited bandwidth for some time to come. For mobile users, access to applications will take place over bandwidth-restricted connections, precisely the environment in which thin-client computing excels.

As obvious as this trend may appear, there are many challenges. First, to ensure portability and minimize power consumption, the processing power of handheld devices will be limited. And because bandwidth is constrained, any Java applications designed for this environment must be very compact. Screen size will also be limited to accommodate a small form factor. In the short term, expect to see custom HTML-oriented applications emerge as the most common application platform for these devices.

Maybe there is a viable wrist-watch thin client in the future, but you'd probably need special magnification goggles to use it. Meanwhile, plenty of thinking remains to be done about the most appropriate environments for thin clients and how to reengineer applications to take advantage of this approach to client/server computing. Server vendors will love the spike in revenue as thin-client computing goes mainstream. And IT managers will love the stability and reduced costs associated with this new twist on an old theme.

Send your comments on this article to Dave Molta at dmolta@nwc.com.