As Sun prepares to release its first Intel-based Linux servers in August, some analysts are arguing that the company should concentrate on developing software - and should stop developing its own UltraSparc processor range.

Sun's forthcoming Intel-based server is codenamed Big Bear, though it is expected to be renamed as the Cobalt LX50. It will move Sun closer to the product portfolios of its major rivals, IBM and HP. This will not be the first Sun product with Intel chips. The company produced Intel-based workstations some years ago, and has a connection with the processor family through its purchase of Cobalt last year.

The key difference is that the LX50 is expected to be launched with Linux as its operating system, rather than Sun's Solaris. Earlier this year, Sun stopped supporting Solaris on Intel.

The LX50 is being targeted at the mid-range market, where Intel-based systems running Windows and Linux are becoming firmly established. However, as Tony Lock of analyst firm Bloor Research has noted, Intel-based systems are becoming increasingly suitable for the kind of high-end roles that Sun's Solaris systems usually perform. "Sun has issues to address with Intel systems running both Linux and Windows," he said. "Unisys and IBM have both created Intel-based servers with the serious systems management characteristics normally found only on larger Unix servers and mainframes. Also, with the launch of Itanium 2 [this month], Intel systems are becoming more powerful, so both Windows and Linux are now more of a direct competitor."

He said there is a danger that Sun could find itself squeezed. "It is being pushed down at the top by IBM and HP with their versions of Unix, and pushed up from the bottom by Windows and Linux - with IBM and HP very prominent there as well," he said. "In addition, Sun is smaller than its competitors and is still using its own processor technology with UltraSparc, so it has to run its own R&D operation rather than exploit [R&D] work done by Intel."

IBM has said it is still committed to its own Power processor family, though some analysts have asked whether the cost of supporting in-house R&D for the processors will be justified. By contrast, HP has abandoned development of its own PA-Risc chip, just as the recently acquired Compaq ditched the Alpha processor line.

So is Big Bear a pointer to a future strategy for Sun, away from developing its own processors? This is certainly the view of some analysts.

"Sun has some good software technology, and with Solaris it has one of the de facto standards against which software vendors develop applications," said Lock. "With UltraSparc it is in perpetual battle with Intel and IBM. What makes matters worse is that it sold a lot of hardware to the now failed dot-coms, so there are fewer upgrade sales possible and a lot of good second-hand Sun hardware available at low cost. Under this pressure, it could choose to become a software company."

Phil Dawson of analyst company Meta Group agreed. "The future for Sun is to earn money as a software company," he said. "However, the company is geared for hardware. I am happy that the UltraSparc roadmap will deliver, but performance is now going down the list of users' priorities. The days of the big Sun server running Oracle are going."

Bill Claybrook of analyst firm The Aberdeen Group said, "For the time being Solaris is still the main ticket for revenue for Sun, though as Linux grows up and scales more readily it will start to compete with Solaris. For now, though, Linux only scales to four-way systems, with eight-way coming. Solaris scales to 100-plus."

However, Claybrook said he is not so sure that Sun will, or should, move away from hardware development. "It has a roadmap for UltraSparc development, and the company sees a long-term future for the processor and Solaris. I don't disagree with that," said Claybrook.

Have your say: contact IT Week