Early gravedigging surrounds HP solutions
April 21, 2010
But such forecasts are not heard anymore about the HP 3000. HP's favorite replacement, the Itanium-based Integrity, draws such dire predictions these days. Itanium may deserve them just as little as the HP 3000 did, at least until Hewlett-Packard announced its exit from the market in 2001.
Doubts about long-term Itanium success are not difficult to locate. Although the most dour analysis comes from the industry-wide The Register IT Web site, developers and founding partners in the 3000 community see declining prospects for Itanium, too. Some doubts might be based upon phantom chills from HP's 3000 pullout. One developer who has customers on the 3000, with others migrating, said HP ought to be engineering ahead for the inevitable.
"I would be more interested to hear HP quietly assembling a group of engineers to get HP-UX to run on the Intel/AMD true commodity server CPU -- Xeon/Opteron -- and finally admit Itanium is a bust," he said. Few HP partners want to go on record with their skepticism, but their development dollars make louder statements. Some are using Dell servers and VMware to host HP-UX in emulation.
But this chip has been tarred for more than five years. Despite those doubts, HP still swears by the only processor which can run HP-UX. Martin Fink, the VP of HP's Business Critical Systems unit, most recently testified when the Tukwila Itanium generation rolled out.
I have turned the spade of Itanium gravedigging myself several times over the past two years. That's been a period of delays and redesigns to build a successor to the current Montecito generation of the chip. In early March the chip's creator Intel and its chief customer HP touted the new Tukwila 9300 generation. HP trimmed up a 20-minute Webcast into 2.5 minutes of Fink's headlines, preceded by the VP of Intel's Architecture Group Kirk Skaugen.
"More than 80 of the top 100 companies in the world have deployed Itanium for their most mission-critical environments," Fink said. He added that HP's implementation of the chip, which employs a subsystem only HP can provide, makes Itanium a success among such customers.
"There's storage, there's networking, there's managability, and integration capabilities. Now take a step back and think, who's been building servers for 30 years? Who's been doing networking for 30 years, and who has the No. 1 and No. 2 position in storage, internal and external, and who's been doing IT for decades"
Of course, to go back 30 years would point to HP's success with the 3000. One supplier of migration and data management solutions for the 3000 notes that such fully-integrated systems offer a stable, superior environment. Birket Foster of MB Foster says that people who focus on a chipset alone miss the complete picture.
"There's a lot of things that are more important than speed these days," he said. "Some of those are things HP has built in for virtualization, and others are the things like digital signal processing and caching that HP anticipated when they first began this design." (It's easy to forget that Itanium started in HP's labs in the late 1980s, engineered until the company chose to pass on building its own fabrication facility.)
Foster pointed out that HP took the main Itanium CPU and built chipsets. He said he's seen evidence of companies employing Itanium for white-box systems, as well as vendor-labeled enterprise servers.
And industry adoption and critical mass in the marketplace? Foster said that customers seem to be willing to let the chips fall where they may. Intel's Skaugen said that 85 percent of the world's Itanium systems are HP's, running HP-UX. Groupe Bull in France and Hitachi enterprise systems also employ the chip, but the growth ramp-up is still in the future, according to Fink.
Foster said HP-UX and Itanium look like safe choices for enterprise-grade customers, rather than small and medium-sized businesses. And the adoption pattern for migrations bears out his analysis: the overwhelming majority of customers reporting on their 3000 migrations -- small companies, many of them -- are choosing Windows and the Xeon/Nehalem chipsets. But Itanium has more improvements to come.
"They're going to do chip releases every two years," Foster said, referring to the Paulson and Kittson generations. "In every company there are going to be people who are detractors of a technology. While there are multiple, different lines of products, people are always going to say their choice is better. Or at least trash the other one, so it leaves it open for people to pick something else."
Guy Smith, the founder of marketing services firm Silicon Strategies and a name familiar to readers of the NewsWire, posted a message about the continued slide of Itanium, in 2005. "Most of my rants are merely showing how it is not winning in the marketplace," he said yesterday looking back on that post, "when defining 'winning' as outselling alternatives." He said nearly five years ago
Any time news starts to sound like a Monty Python script, you know the end is near. [InfoWorld's 2005] headline reads "Itanium: not dead yet," which is so close to a scene from Monty Python's Holy Grail that one has to laugh. But the article turns sad by paragraph 3, and shows in this [Itanium] Alliance the same insanely hopeful straw-grasping that people display when loved ones are about to become the dearly
departed -- fanciful fits of abject hopefulness.
The timing of a demise is subject to much estimation, Smith adds today. "We write obits long before the corpse is cold," he said. "Many people predicted the death of COBOL in the 1980s, yet [IT consultant and 3000 volunteer] Bob Karlin will give you an earful about how it is both a programming language and the undead."
"Itanium may indeed be the best there is and ever was," he said. "So was Betamax for video. The market turned its back on both." Detractors and defenders of the Integrity's fundamental chip may both be right, he added.