Microsoft's Network Wares Still Aren't ConnectingAmy Cortese and Richard Brandt
When it was introduced a year ago, Windows NT was supposed to lead Microsoft Corp.'s triumphant advance beyond the desktop. NT, for New Technology, would extend Microsoft's software empire to "servers," the powerful computers at the heart of the client-server networks increasingly replacing mainframes and minicomputers. After five years of work and $100 million in development costs, Bill Gates was so sure NT would be a hit that he predicted it would easily sell 1 million copies in the first 12 months. Today, analysts put the number closer to 300,000 and say only a third are on servers, Gates's strategic target.
Microsoft's frustrations in networking are nothing new. Its LAN Manager, introduced in 1989, has done poorly against Novell Inc., whose NetWare has 85% of the market. Similarly, Windows for Workgroups, a program for sharing files among groups of workers, has not slowed Notes, Lotus Development Corp.'s "groupware" hit.
But the NT setback could have wider repercussions. As Gates's ambitions have grown, so has his bet on NT. Now, NT is expected to be the foundation for a slew of advanced systems, including Tiger video servers, and Cairo, an "intelligent" operating system that will, on its own, fetch information on a network. NT is also the base technology for new server applications--programs for electronic messaging, groupware, and network management. In short, NT is fundamental to Gates's plan for making Microsoft a communications giant.
What went wrong? For one, even though Microsoft pitched NT for running large networks--even whole companies--the software lacked many of the features and applications to do so, and it didn't work well with existing computer setups, such as those running NetWare. NT also requires lots of computer memory to do even simple tasks, and like many new software products, it still has a few bugs to be worked out. "NT turned out to be bigger and uglier and slower than anybody had expected," says Stan Hanks, president of NetSmiths Ltd., an Edison (N.J.) computer consultant. Meanwhile, sales of Unix, a mature operating system, are booming in the server market.
RIGHT STUFF. The task of making NT live up to its billing falls to James E. Allchin, vice-president of Advanced Systems. Gates persuaded Allchin, chief technologist at network software maker Banyan Systems, to join Microsoft in 1990. A computer scientist who held most PC software in low esteem, Allchin was reluctant at first. Gates kept pushing because he figured that Allchin, 41, had the right stuff: As a PhD candidate at Georgia Tech in the early 1980s, he had written a program called Clouds that linked several minicomputers together to act like one huge machine. That could come in handy in pulling off all the communications tricks Gates has in mind.
The comeback plan for NT is already in gear. The second release, code-named Daytona, is expected this summer. In addition to cleaning up bugs, Daytona will require less memory and will work with more systems, including NetWare. And Microsoft has been persuading makers of Unix and mainframe applications, such as Computer Associates International Inc. and Oracle Corp., to write programs for NT. Microsoft's own NT programs, such as "Touchdown," a groupware offering, and a revamped SQL Server database, are imminent.
Meanwhile, momentum for NT is starting to build. PaineWebber Inc. analyst Michael Kwatinetz predicts that as many as 500,000 copies of Daytona could sell this year. Still, it won't be until 1995 or 1996 until NT really catches on, he says. "The thing you have to admire about Bill Gates," says Kwatinetz, "is that he understands that making technology prevalent is not a short-term process." Good thing.