Taking Down NT on Alpha
I just couldn't keep quiet about Compaq dumping NT on Alpha.
After all, I introduced Alpha to many of Digital Equipment Corp.'s major customers in Minnesota, starting in the spring of 1990. I was there when the engineering team unveiled the prototypes for the first workstations. And I was there during many of the early meetings in New Hampshire when different engineering and product groups presented their plans for the next several years.
I knew some of the engineers who were at the chip conference in early 1992, when Digital introduced the first Alpha chip. They stole the show, and I don’t recall anything before or since that caused such a stir. Alpha was going to be a revolution in computing and nobody else had anything that could come close.
For years, I carried around a little Alpha chip laminated into a business card. After a while, it got so old that little metal flakes ground off the top of the chip and the plastic laminated edges of the card became yellow and dog-eared. But I still pulled out that card to show people because I was proud to play a small part in the beginning.
When the news hit that Compaq dumped NT on Alpha and laid off the engineering staff on the West Coast, and that Microsoft dumped 64-bit NT on Alpha a few days later, it was a bombshell and an emotional shock to me even after all these years. I decided to wait a few weeks for the shock to wear off and then see if I had anything intelligent to add to the discussion.
Here are a few personal observations.
NT on Alpha never performed well at the desktop. I tried a couple benchmarks a few years ago and found that lower-cost Intel machines performed at least as well and often better than Alpha machines for common PC applications. Evidently, some compute-intensive high-end applications built especially for the Alpha environment did work well, but I never saw any first-hand. So that raised the obvious question: Why would most people spend the premium to buy an Alpha NT workstation?
A few years ago, when I had some cash burning a hole in my pocket, I bought a couple Alphaservers. In fact, Bertha -- an Alphaserver 2000 -- is still the primary domain controller for the SCOTTCONSULTING -- soon to be renamed CROSSCG -- Windows NT Domain.
Every component inside Bertha is way more expensive then the equivalent Intel component. For example, consider the DEC KZPSA board, which is a PCI-to-Fast Wide Differential SCSI controller. Its retail price was around $1,500. The equivalent Adaptec controller at the time, an Adaptec AHA 2944W, cost about $300. Other components and peripheral devices have similar economics.
One of my other Alphaservers, Bessie, an Alphaserver 1000, recently had a hardware problem. We use Bessie as a test machine. Sometimes it runs Tru64, sometimes OpenVMS, sometimes NT. About a month after its three-year warranty expired, Bessie forgot all its firmware settings and refused to finish its power-up cycle. Our best guess was that its CMOS battery died. No problem, right? Just replace the battery. After looking at the motherboard, we found the battery is really a Dallas chip soldered to the motherboard. This means we need a whole new motherboard to replace a $5 battery. I’m afraid to even look for a replacement because I know it’s going to cost a fortune.
NT was supposed to be Digital's -- now part of Compaq Computer Corp. -- volume outlet for Alpha. This market was supposed to keep chip manufacturing costs in line so Compaq could keep developing Tru64 and OpenVMS at the high end. Since Compaq says Alpha accounted for less than 2 percent of their NT sales, I have to believe they never came close to the volume they wanted. So without a high-volume outlet, what happens to the rest of the Alpha market and ongoing engineering and development? Does anybody seriously believe Linux will generate significant Alpha volume? Why pay the premium over Intel? And does anyone believe all the reassuring statements of commitment coming from top Compaq executives?
I share a problem with many other customers who bet on Alpha. I now have an Alpha NT machine that will never support Windows 2000 or any flavor of 64-bit NT. This means I must migrate to an Intel server if I want to stay current with technology. I read that Compaq is offering free Tru64 and OpenVMS licenses and sweet upgrade deals for all of us stuck with Alpha NT machines. I’d love for somebody from Compaq to call me and offer a deal. --Greg Scott, Microsoft Certified Systems Engineer (MCSE), can be contacted at firstname.lastname@example.org.