Paul -
Agree on Spencer the Katt - kind of the national inquirer of the trade press ('Aliens take over Senate!'). But let's look at just what CPQ might do with Alpha, and what a smart play for Intel might be, before we assume dismiss the issue. The next month or two should tell the story IMO. But isn't CPQ still Intel's largest customer? I think your opinion about Alpha is a pretty accurate reflection of attitudes around Intel, both inside and outside the company. Maybe you are right, and Alpha will just go away. But maybe not.
CPQ has more than 35% of the Intel-based server market, more than the next 2 vendors combined. Every server and professional workstation product CPQ makes is made, and has always been made, with Intel processor technology.
DEC, despite its weak market presence, has had a strong relationship on both the engineering and marketing sides with Redmond. They are the lead partner in Exchange installations, with more than 2 million seats. CPQ is second, with more than a million.
The quickest path to better performance for MSFT is not SMP - it is faster individual CPU performance and large memory support. MSFT has been very reluctant to provide more than token support for the 36 bit memory architecture - they want to go right to 64 bit since they only want to change core OS components once. Right now, Alpha gives MSFT both VLM and faster individual CPU performance.
MSFT is developing all of the core OS components on Alpha. The Merced simulator does not provide them with the ability to test memory management, driver support, allocation and protection schemes. The delay in Merced silicon means that 64 bit NT will be developed, debugged and deployed to developers on Alpha.
So conditions now are different than they have been at any other time in Alpha's history. MSFT is clearly doing both 64 bit work and advanced OS development on Alpha as the primary platform. This is no longer a case of MSFT offering 'another' supported processor. Alpha will be THE ONLY supported processor for 64 bit NT development. And please don't tell me that ISVs will develop on the simulator when they can have real hardware.
Microsoft's key enterprise penetration application, Exchange, is being driven by CPQ / DEC in about 1 out of 2 installations. Early Alpha designs which lacked key features (like byte manipulation) have been flushed out of the customer base, so MSFT can write much more efficiently for native Alpha than in the past.
I think CPQ was on a path to use Alpha as an on-ramp to Merced dominance until they began to see that Merced would be a lot later than they wanted. That prompted a change from Alpha as a 64 bit transition tool to Alpha as an alternate product strategy. A few heavy handed moves from Intel and it could become Alpha the high-end processor strategy.
I don't believe CPQ is 'desperately' looking for help from MSFT - I think they are just keeping in close contact with their OS partner as their strategy shifts toward Alpha as more than an interim step.
What if the press has it wrong, and it is really MSFT pressing CPQ to provide more support for Alpha? What if the strategy was laid out before CPQ announced the merger, and what we are seeing is the implementation of a plan that has been in the works for 6 months or more? This is not Packard-Bell or Aunt Nellie's white box company, this is a company that dominates the market for which Merced is intended. They will have years to execute and consolidate their position before their rivals can get on the playing field.
Your attitude reminds me a little of IBM's attitude about the early clone vendors in 1983. Instead of looking at all the ways that someone might succeed in competition for IBM's PC business (a business they had nearly 90% of at that point), IBM spent time looking at all of the reasons why the clones could not succeed. They did almost nothing proactive to protect the future of their PC business. They instead concentrated on their internal roadmaps, ignored the clones, and over the next 4 years went from a 90% share to a 30% share of the business they created. Look at where they are now, struggling to maintain a 2 digit market share in the PC business.
If Merced is a key future product for Intel, I would think that they would pay attention to what the largest systems partner for Merced is doing in 64 bit development. |
