Sunset for SPARC?
It is no accident that Intel’s promotional material that accompanied the Itanium’s release focused exclusively on comparisons to Sun Microsystems’s UltraSPARC-II and UltraSPARC-III line of processors. Besides having the only processors without stronger SPECint2k performance than Itanium, Sun is also the only major system vendor that has no plans to introduce an IA64 based product line in parallel with its house brand RISC systems. So Intel can draw enormous cross-hairs over the successful Unix system vendor and its iconoclastic CEO, Scott McNealy, and show off its new 64 bit offering in a favorable light without the ticklish problem of offending a competitor that is also an important customer. It may have also picked the juiciest target, one ripe for the plucking.
There are a remarkable number of similarities between Sun Microsystems today and Digital Equipment Corporation (DEC) in the second half of the 1980’s. At that time, DEC was enjoying an incredible run of profitability and growth on the back of the VAX, a broad family of computers ranging from several thousand dollar microprocessor-based low end workstations to million dollar plus mainframe-like servers equipped with multiple processors built from discrete standard and custom ECL components. Like SPARC, VAX implementations certainly weren’t the fastest computer in their class. There were competitors such as Perkin-Elmer, Gould, and Data General that offered machines that were faster, cheaper, or both.
What DEC had going for it was a sterling reputation as a reliable supplier, a robust and powerful operating system (VMS), and an industry leading ability to easily configure and implement complex networks of VAX systems and grow them as needed. This capability made the VAX really stand out from its competitors and led to rapid sales, which in turn attracted software developers. The growing software base in turn reinforced the VAX’s growing dominance in the mid range computer market. Sun has essentially followed the same path in establishing SPARC as the highest volume server class RISC processor. The tide finally turned against VAX when it became apparent that DEC just couldn’t keep it competitive in cost and performance with the rapidly growing field of RISC processor based systems (including Sun’s SPARC). By the early 1990’s, the VAX was in steep decline and the end came on September 30, 2000 when Compaq officially took the last order for VAX systems.
Will IA64 do to SPARC what the RISC competitors did to the VAX? The primary argument against this analogy is the simple and compelling fact that at the processor level Sun has trailed its RISC competitors in performance for more than a decade, yet it has enjoyed enormous commercial success. The critical difference now may be Intel’s ability to duplicate the devastatingly effective Xeon business model for IA64. The economic advantages of selling standardized, high volume server components through multiple system vendors has allowed the lowly x86 to grab a huge chunk of the entry and mid level server markets and workstation market from both Sun and technically superior RISC players. Software seems to be the only factor working in Sun’s favor. Besides the obvious and glaring disparity between the quantity and range of applications available for SPARC and IA64, Sun’s house brand of Unix, called Solaris, enjoys an enviable reputation for functionality and robustness. But then again, so did VMS (which ironically is enjoying a mini-renaissance on the Alpha. EV6, Extended VAX indeed!).
Discuss (15 comments)