April 22, 1997 2:00 PM PDT
Superservers near death
The last nail will be hammered into the coffin next week when NetFrame Systems (NETF), the last vendor of mid-range Intel-based server systems with proprietary features, revamps its servers to use industry standard technologies, like everybody else.
NetFrame's new sales and marketing strategy is to get its share of a higher-volume market by eschewing the reliability features based on proprietary technology, features that its customers have come to depend on.
"There is really no future in selling a low-volume superserver," admitted Steve Huey, vice president of marketing for NetFrame. "We were the last of the superserver companies. I don't think the concept plays anymore."
The term "superserver" was coined in the early 1990s to describe products from NetFrame, Tricord Systems (TRCD), and others that featured hardware technology for increased reliability, such as secondary processors to do data crunching or take over when one processor happens to fail. These technologies made the boxes an attractive alternative to more expensive servers running Unix operating systems, and industry observers at the time predicted a bright future for the superserver market.
But over the years, customers decided that they wanted to use servers based on de facto industry standards like Intel processors or Unix operating systems. The superserver's demise was brought about by the advent of high-speed, low-cost Intel chips such as the Pentium Pro, the increased popularity of four-way multiprocessing servers, and an increase of reliable industry-standard components.
Other vendors of Intel-based servers already dwarf the scalability of a NetFrame system. Companies such as Sequent Computer Systems and Stratus Computer offer machines that are really alternatives to mainframes but run some variation of a Unix operating system.
Analysts say that Unix vendors should pay attention to the lessons of the superserver companies, especially as lower-priced Pentium Pro models become more popular. Sun Microsystems, for example, appears to have already gotten the message; later this year, the company plans to introduce a series of workgroup servers that use its own SPARC microprocessors at price points comparable to Intel-based systems.
In the meantime, the companies who once led the superserver charge are left to limp along into new markets, if they can.
Although NetFrame is ready to change its tune now, it has already paid a heavy price for its loyalty to the superserver philosophy. The company reported a loss of more than $11 million and a 32 percent drop in revenues for its most recent quarter compared with the year before. It has already lost more than $8 million in 1995 after enjoying relatively strong profits from 1992 through 1994.
NetFrame executives said they now plan to offer systems comparable to those on the market from Compaq Computer, bringing price points down from the $60,000 range to about $25,000. The company will also certify components such as storage systems to insure that they continue to offer reliability features to customers. NetFrame will also continue to offer its fault-tolerant technology as an add-on component for an extra fee.
"It's clearly a move to grow the revenue of the company and get it back to a profitable position," NetFrame's Huey said. "I don't think we can continue to have the mentality that we can charge more."
But some market analysts feel the company waited too long. "Their days are numbered," predicted Peter Lowber, analyst for the Datapro Information Services Group.
Tricord Systems, who once competed fiercely with NetFrame, has gotten out of the server business altogether, focusing its efforts instead on software that increases the data throughput of server systems.
Tricord was forced to change course in February after it reported a $15 million loss for fiscal 1996. Company representatives said they had to choose between software and hardware development, and the software side of the business ran with less overhead expenses.
Five years ago, the superserver market seemed like a good bet to many, yet analysts say that NetFrame and Tricord could have seen the end coming. "I think the market has said for a long time that it doesn't like proprietary technology, especially when there is a standard alternative," said Bill Hargrove, industry analyst for Currid and Company. "I can't say I'm too surprised."