Identity Crisis Jolts Workstation World.

Old order changeth as Wintel muscles in

Knowing what constituted a workstation used to be easier. In the first place, the vendor told you what it was. Where that approach didn't satisfy, an integrator could look to the underlying technology or the system's competence in graphics or, perhaps, memory. Failing this, the application frequently gave the game away. If the system was used in CAD/CAM or, perhaps, EDA, it was likely a workstation.

From a strictly technological point of view, it was usually safe to identify any RISC-based desktop system that ran Unix as a workstation, but the growth of the Intel-based, NT-dependent systems undercuts even this basic distinction. NT is dominating shipments right now, so the IA series products are assuming the mantle of workstation-dom.

Part of the challenge comes from the fact that workstation technologies are in a state of convergence. Users and even integrators will disagree with where lines are drawn between PCs, servers, thin clients, and workstations. Another challenge is the tsunami that is the network. If the "network is the computer," where does the workstation stand?

Workstation Or PC?

One place they stand is as a profitable market. The overall workstation market in 1999 came to a $9.8 billion split among several leading vendors (Fig 1). The top three vendors of workstations, Sun, Hewlett Packard, and IBM, hold about 60% of the revenue. However, Dell has doubled its market share since 1998 Sun holds the lead in the RISC/Unix space with Dell holding the lead in the Intel Architecture/NT product lines.

Another advantage was the opportunity to create a workstation branding group for scalability ... buy the workstation power you need and no more. While this constitutes intelligent branding strategy, it allows marketplace confusion between identifying the workstation versus the PC. The Compaq Deskpro and many of the HP "Kayak" series workstations share the attributes of the workstation and the PC, as well.

The greying of the workstation began roughly five years ago, as PC manufacturers embraced workstations as Intel-based systems running the NT OS. It gave the PC maker a foothold in a new world . . . high performance computing. One of the advantages of the new world was a pricing structure that offered more margin.

The creation of lower end workstations produces an anomaly that analyst Pia Rieppo at Gartner/Dataquest calls a "volume trap." She points out that many workstations are now technologically indistinguishable from the PC to the point that some so-called PCs outperform so-called workstations.

Rieppo recounts the situation last March, where Intel rushed its gigahertz processor to market to counter the introduction by Advanced Micro Devices. These processors were targeted to the high-end consumer game space. This stymied the workstation product management, who could not offer the latest in workstations . ... the latest went to gaming PCs. The upcoming Pentium 4 illustrates the volume trap, as well. P4 is targeted at consumer communities when it is introduced. The dual-processor chipset aimed at the workstation vendors such as Compaq, Dell, HP, Silicon Graphics, and IBM will not be released until six months after the PC-based P4.

Bundling Is The Future

It seems likely that a pure hardware/OS sell will be seen by end users as a commodity sell. This is unfortunate, since top technological developments like the Alpha architecture (owned by Compaq through their acquisition of DEC assets) could be less important than smart marketing.

Yet Dataquest's Rieppo notes that in about five years, the important workstation features will not be technology or marketing, but the offering of bundled elements, including applications and services. This pushes the workstation vendors to compete based on the usefulness of the entire turnkey solution. Conventional business advantages such as effective vendor/user partnerships, as well as prompt and careful time to market would be the foundation for workstation's future.

While the vendors of workstation technology are planning, what do the buyers have in mind? Those who plan to stick to the RISC/Unix space will have to budget more for their systems with 60% of their systems costing above $5,000. Applications are king in this space, where users will buy according to task requirements. The Intel Architecture/NT devices will appeal to cost-conscious buyers whose budgets outweigh the applications requirements. The average system here is likely to cost under $3,000, reminiscent of server and thin client pricing.

OS flu Jour

Although technology distinctions will yield pride of place to applications and bundled solutions, the workstation community must still follow the popular operating systems. In a recent end-user survey, Gartner/Dataquest asked which OS they would choose for their IA-64 based workstations. The result is illustrated in Fig 2.

A significant trend was the exceptional showing by Linux. There is little doubt that the popularity of the open systems philosophy is responsible, especially when supported by cost advantages and ease of use.

The history of computing evolution has always had its elements of doubt. Will this processor survive? Will this operating system go the distance? Yet the future of the workstation market is not so much obsolescence, as it is end-user confusion. If the lines between workstations, PCs, servers, and thin clients continue to blur, users will start buying PCs with bundled solutions. This would drop revenues in the workstation segment, like Newton's apple. At this point, in the workstation's evolution, PC-type devices can be billed as workstations. Absent is a new and meaningful effort at product differentiation; workstations may end up booked and billed as high-end PCs.

"But first your memory we'll jog,

And say a cat is not a dog."

T.S. Eliot