wbmw, Sun - this is an intriguing development, although some of what you posted misses the mark:
Maybe Intel's EM64T announcement destroyed any advantage that Sun thought they would have against the other vendors.
I should point out that Sun's stated strategy with Opteron from the beginning has been to expand their less expensive offerings and build volume. Current reported efforts by Sun to port Solaris to Itanium and Power do not conflict at all with their public position with regards to Opteron.
However, it does have interesting implications:
1. Some of us, including me, have speculated that Sun will work with AMD on K9 and future architectures to grow Opteron beyond their lower end offerings. This news shows there is no special arrangement there. :(
2. The SPARC folks must be sweating bullets around now. Sun spoke highly of evolving SPARC into a massively parrallel high end architecture, this news is contrary to that position.
3. And this should definitely scare you if you're a SUNW investor.
I agree! (Luckily, I do not hold SUNW; I do hold MSFT which had much, much better news this week. :)
So what it Sun's high end strategy? SPARC? Itanium? Power? Are they trying to out-IBM IBM by offering all this stuff - now there is a recipe for disaster!
If I were in charge at Sun, I'd guide a certain amount of dynamic competition to determine the best strategy for high end computing. I would not take a shotgun approach of three high end architectures, as implied today. I would support pilot projects to see what the single focussed strategy should entail.
Supporting Solaris on four architectures will split their market into too many tiny segments, cause irritating incompatibilities, make too many configurations to test affordably and result in lots of finger pointing between the OS folks and the hardware folks when bugs appear. What does a hardware person say when an important bug appears on their platform? "Does it reproduce on the other processors?" If not then the hardware folks wind up chasing a problem which may very well be in the OS but only appear on their particular configuration. If it does appear on other processors the software folks chase problems which may be in a common hardware component between systems. Lots of expensive hours go by as each group (already busy) passes the buck.
A real mess.
Other than that, it is a great idea!
I was considering purchasing Sun stock a while ago, but decided to hold off. Now I am very glad I did! They are ping-ponging between different answers to their deteriorating market position. It is past time for them to focus on a single strategy and do it well!
I will note that they did not make any Power or Itanium product commitments, merely acknowleged that they are working on these Solaris ports. It may never see the light of day. I guess they are running the idea up the flag pole, and watching to see if anyone salutes.
One-fingered salutes don't count.
AI
