Paul DeMone (pdemone@igs.net) on 8/30/06 wrote:
---------------------------
>Peter (Peter.Huisken@OneTwoMove.nl) on 8/30/06 wrote:
>---------------------------
>>Paul DeMone (pdemone@igs.net) on 8/29/06 wrote:
>>---------------------------
>>>Perhaps. Nevertheless Linux on POWER is a major
>>>push for IBM. IMO the spectacular preference of IBM
>>>customers for its x455 IPF server over POWER for
>>>Linux was a major reason for IBM killing its IPF line.
>>
>>IMO, this is nonsense. The idea of killing a successful
>>product because it is selling too well is very special.
>>Besides, having IA64 gear gives the sales dept access to
>>potential IA64 customers.
>>
>
>IBM has a long history of knifing good solutions that
>can offer higher value to customers but at the cost
>of existing high margin IBM products. The x455 is
>unusual only in that IBM's usually adroit sales and
>marketing braintrust didn't foresee its non-x86 Linux
>customers overwhelmingly prefer IPF to POWER and
>let the embarrassment build to the point IDC reported
>this to all and sundry. That made a public execution
>inevitable. The fact that IBM still sold enough x455s
>in the first 3 quarters of 2005 to be 4th on the list of
>IPF OEMs, long after its announcement it was exiting
>the IPF market, shows that IBM's claim of little interest
>among its customers was BS. IBM forced its customers
>to decide between staying with IBM and going with Linux
>on POWER or elsewhere, probably HP, for Linux on IPF.
>The reference to IDC data I just unearthed shows what
>a dangerous gamble that turned out to be. ;-)
>
>BTW, why don't you do research on why IBM buried its
>801 RISC research project for a decade. That is the
>classic example of IBM shelving new and cost effective
>technology in order to protect its high margin cash cows.
>Even when IBM was forced to embrace RISC its miser-
>able first effort, the PC/RT, demonstrates an extremely
>conflicted approach, likely held back by internal vested
>interests in the status quo.

IBM have gone on record as saying that IPF just did not sell well enough for them, those IPF Linux sales being a pittance in the general scheme of things. If they were that bothered about commodity boxes selling Linux they wouldn't bother with Opterons and Xeons the latter in 32 socket configurations as they are far more a threat to Power than IPF. In fact they have gone on record* as saying they price their Power boxes compared to Opteron boxes so they are obviously embracing and countering commodity competition against Power. Frankly this pet theory of yours does not sound any more plausible no matter how many times you repeat it.

*
"We take Sun and HP Opteron machines as serious competition," explains Karl Freund, vice president of product marketing for IBM's System p5 division within its Systems and Technology Group. "We cannot charge a premium for our performance. So these machines are priced right on top of Opteron-based machines. If our pricing is more than $2 off the price of an Opteron machine, my team still has work to do."

http://www.itjungle.com/breaking/bn082306-story02.html