In my business -- analyzing and forecasting market and technology trends -- it usually doesn't pay to admit you were wrong. Usually, you want to remind people what you got right and count on the miscues having a short shelf life.
Actually, my misforecast is revealing of a critical industry phenomenon. Last year IDC miscalculated the effects of commodified technologies. We were expecting Microsoft's Windows NT to have a more damaging effect on the workstation market than it did, while expecting NT and other shrink-wrapped operating environments to cannibalize more of the traditional Unix systems space.
Well, those trends "underperformed our expectations," as we say in the trade. In fact, the top five workstation players combined (Sun, HP, Digital, IBM, and SGI) grew 20 percent in shipments from 1993 to 1994, almost twice as fast as we were forecasting. And turnkey Unix servers from those companies, plus AT&T and Data General, grew about 40 percent. That compares to shrink-wrapped versions of Unix from the Santa Cruz Operation (SCO) and Novell's UnixWare group, which grew less than 10 percent combined.
Don't get me wrong. NT workstations totaled about 200,000 units -- better than any single workstation OS except SunSoft's Solaris. NT servers did 80,000 or so, and SCO did more than 150,000 server licenses itself. Each of those is better than any single Unix flavor (and way below NetWare's 580,000 servers -- but most of those are used for simple file sharing).
For the last few years, one of IDC's biggest schticks has been the notion of computer industry "disintegration." That is, competition at every point of the value chain means that building your system from the processor to the software to the distribution channel won't buy you much. Each technology or business has had to be competitive in its own right.
Building an advanced system -- one that runs complex, critical applications like design engineering, currency trading, or inventory management -- out of an operating system from one supplier and commodity hardware to another has been a natural fallout of this trend.
But one corollary to the disintegration theory is that when you're in the early stages of a new technology adoption or business practice evolution, the components need to be a little more integrated. Maybe it's still a little early in the life of distributed, client/server computing
Assuming the risk
Unix-based advanced systems suppliers have exploited their opportunity to prove the strength of their applications and database servers, positioning their overall stability and scalability -- as well as their rich development and management environments -- against NetWare and NT servers. The same is true for the heads-down applications for which people use Unix workstations.
Equally important is the fact that the mechanism for delivering and supporting these advanced systems is still primarily through direct sales. Even if Sun moves half its workstations indirectly, there's a sophisticated sales person from its VAR or ISV holding the customer's hand.
Microsoft, Novell, and SCO still have to attract the appropriate partners to take on the customers' risk in integrating the disintegrated components. That, or shrink-wrap the whole process, but that's the topic for another column.
About the author
David Card is a director of the Software Research Group at International Data Corp., located in Framingham, MA. He can be reached at firstname.lastname@example.org.
If you have problems with this magazine, contact email@example.com
Last updated: 1 May 1995