HP: last Itanium man standing
Nehalem lives the dream
Comment Make no mistake: If Hewlett-Packard had not coerced chip maker Intel into making Itanium into something it never should have been, the point we have come to in the history of the server business would have got here a hell of a lot sooner than it has. But the flip side is that a whole slew of chip innovation outside of Intel might never have happened.
In this respect - and some might even be so cynical as to argue in only this respect - Intel's and HP's troubled chip development marriage and its resulting Itanium love child can be deemed a success. Without the threat of Itanium, which was never really fulfilled, perhaps IBM would have never knuckled down and put some money into decent Power chip development, which allowed the company to go from joke to dominance in the Unix server racket.
And without Intel relegating 64-bit processing to Itaniums and leaving Xeons to 32-bits, there would not have been a gap in which Advanced Micro Devices could leap and create the Opterons, which are the inspiration for the Nehalem family of processors that have put Intel back in the driver's seat when it comes to server CPUs and which give Intel a chance to have the kind of dominance in the data center that it enjoys on the desktop in the not-too-distant future.
None of us has the energy or the time to go over the multitude of sins that Intel and HP committed with the Itanium, ranging from the hubris of changing the instruction set to the stupidity of having too aggressive a deliver schedule in the early years to all but ignoring Itanium in the later years. But to understand what will happen to Itanium - and what will not happen to it - we have to review a little history.
The whole Itanium plan was predicated on all the major server vendors porting their platforms to the operating system, and as the 1990s came to a close and Itanium was still a threat rather than a disappointment, all the major OS makers swore their fealty to Itanium. That includes IBM with AIX, Sun Microsystems with Solaris, Hewlett-Packard with HP-UX, Santa Cruz Operation with OpenServer, Compaq with OpenVMS and Tru64 Unix, Microsoft with Windows, various emerging Linux players with their revs of that open source platform, and myriad proprietary and mainframe platforms (many of which help prop up the Itanium chip today).
The enthusiasm was more fear than anything else - fear of crossing Intel and suffering the consequences in the volume x86 business and fear of being left out on a big opportunity. And thus early forecasts, which had Itanium server sales kissing $40bn in 2001, look ridiculous now as we look back on them.
With HP finally getting around to launching machines based on the quad-core "Tukwila" Itanium 9300 processors, which made their debut back in early February with only HP and Super Micro committing publicly to using the new chip in systems. While none of the shippers of prior Itanium systems would bad-mouth the Tukwilas, the unwillingness of Unisys, Fujitsu, Silicon Graphics, Bull, NEC, and Hitachi to even admit they were working on Tukwila platforms was an astounding reversal of what server makers were saying back in 1996 and 1997 when it looked like Itanium would take over the world. And maybe Uranus, too.
IBM and Dell pulled the plug on their Itanium lines after only a few years, giving them about as much marketing effort as most politicians push for campaign finance reform. Which makes HP the John McCain of the Itanium world, I suppose, and perhaps a prisoner of war in a concentration camp it helped construct.
But hardware sales are driven by software sales, and software vendors don't write operating systems or application software for chips that don't look like they are going to hit their volumes or provide lots of margin to cover the work, as do mainframes and other high-end proprietary or Unix boxes.
So by the turn of the millennium, the IBM-Compaq-SCO triumvirate that was supposed to get the Monterey/64 converged AIX-OpenServer Unix running on Itanium got the work done and pulled the plug, and similarly Sun Microsystems, which completed an Itanium port of Solaris, sat on it. Microsoft supported Itanium for many years with Windows Server, but the company is always looking for a way to cut back on platforms when it comes to Windows Server. (Remember how Windows Server was supposed to run on x86, MIPS, Alpha, and Power platforms when it was launched in 1994?)
Microsoft relegated the Itanium version of Windows to a database engine with Windows Server 2008, and earlier this month Microsoft said that enough was enough and that Windows Server 2008 R2 was the last release of its operating system that would be supported on Itanium. El Reg broke the story late last year that Red Hat was going to kill off Itanium chip support in the Enterprise Linux 6 distro.
That leaves HP's HP-UX, OpenVMS, and NonStop operating systems, Novell's SUSE Linux Enterprise Server 11, and a handful of proprietary OSes from Europe and Japan on Itanium chips. With the exception of SLES, these customers have no alternatives, any more than IBM and Unisys mainframe or IBM OS/400 shops do. Which means as long as customers have difficulty in moving their applications and as long as Intel is willing to dedicate enough capacity to crank out the 400,000 or so Itanium chips needed (and can make money on the $1.5bn to $2.5bn in estimated annual Itanium chip revenue) to satisfy these customers, then Itanium will be around through the next eight-core "Poulson" and perhaps 16-core "Kittson" Itanium generations.
Assuming Itanium is on a two-year cycle, then there will be Itanium processors available for at least six to seven years, bolstering an Itanium server business that did around $5bn in sales in 2008, according to the Itanium Solutions Alliance. The analysts at Gartner have been cited saying they believe Itanium-based machines comprised 9.3 per cent of total worldwide server revenues in 2009, which works out to $4.07bn. That's an 18.5 per cent decline, which is not too bad compared to an overall server market that fell by 18.3 per cent to $43.1bn worldwide, according to Gartner.
Next page: Niche trap
Why Matt B likes Itanium
Start with a cage containing five monkeys.
Inside the cage, hang a banana on a string and place a set of stairs under it. Before long, a monkey will go to the stairs and start to climb towards the banana. As soon as he touches the stairs, spray all of the other monkeys with cold water.
After a while, another monkey makes an attempt with the same result - all the other monkeys are sprayed with cold water. Pretty soon, when another monkey tries to climb the stairs, the other monkeys will try to prevent it.
Now, put away the cold water. Remove one monkey from the cage and replace it with a new one. The new monkey sees the banana and wants to climb the stairs. To his surprise and horror, all of the other monkeys attack him.
After another attempt and attack, he knows that if he tries to climb the stairs, he will be assaulted.
Next, remove another of the original five monkeys and replace it with a new one. The newcomer goes to the stairs and is attacked. The previous newcomer takes part in the punishment with enthusiasm! Likewise, replace a third original monkey with a new one, then a fourth, then the fifth. Every time the newest monkey takes to the stairs, he is attacked.
Most of the monkeys that are beating him have no idea why they were not permitted to climb the stairs or why they are participating in the beating of the newest monkey.
After replacing all the original monkeys, none of the remaining monkeys have ever been sprayed with cold water. Nevertheless, no monkey ever again approaches the stairs to try for the banana. Why not? Because as far as they know that's the way it's always been done round here.
And that, my friends, is why Matt Bryant still stands up for Itanic
Not that simple
As I remember it, the HP/Intel tie-up looked like a good fit, even though in retrospect I would say that Intel took HP for a ride.
At the time HP decided to jump to a joint-developed chip with Intel, it was engaged in an arms-race with IBM. Tim said that it was the Itanium that made IBM invest in Power, but in actual fact this investment pre-dated the Itanium by at least 5 years. IBM bumped development of Power, PowerPC, RS64, and Power2-7 at various times, but it has been an almost continuous process if we overlook the stumble that happened with the 64 bit PowerPC 620 processor.
The original RIOS based IBM POWER systems, the RISC System/6000, was launched in 1990, and had been under development for at least 5 years before that. The driver was to be an industry leader in the Open Systems market place, as IBM had at last recognized that there was money to be made.
When first announced, the RS/6000 model 530 killed everything on the market stone dead, it was so much faster. HP had PA-RISC running in their MPE/iX line at the time, but it was not a single microprocessor, being built from discrete logic. The RS/6000 caused a huge stir, both because it was so much faster, and also because IBM put significant marketing weight behind their new systems. Sun were immediately knocked off the top of the workstation market, and never managed to really get back up there, and DEC invested heavily to try to produce a really hot chip in the Alpha, that was as a result of the need for speed, was significantly flawed and never really delivered on it's promise.
HP rushed systems to counter the RS/6000 based on the single chip implementation of PA-RISC, and running HP/UX. These were the HP 9000 model 720, 730 and 750, and the race was then on between IBM and HP to see who could have the fastest system. This reached it's peak in the late 1990's, when some models of RS/6000 had marketing lives of less than 6 months.
This was tremendously expensive, and HP, who did not have a big chip-fabrication division valiantly struggled to keep up, but was ultimately doomed to fail.
The way I remember the Itanium being pitched was that Intel were going to take on the development of the PA-RISC single microprocessor replacement, keeping most of the instruction set, but putting in features that would allow the processor to also run x86 binaries, and enhancing the x86 architecture for 64 bit. Intel would get access to HP's IP for the PA-RISC (which included high clock rate silicon and cache IP), and would use their considerable chip making skill to drive the product forward. HP would get a class-leading processor to keep their workstations and servers going. At least that is what was said by Intel.
What actually appeared to happen was that they designed Itanium to be their own processor, with less emphasis on making it a PA-RISC replacement, and more on trying to make it an upgrade path for 32bit x86 servers. They delivered it late, and the product did not live up to their claims as either a PA-RISC replacement, or a 64 bit x86 migration path. Intel attempted to use some of the IP to produce high speed x86 processors, but botched it with the Pentium 4, which was ultimately a dead-end.
Because of the delay, the world in general, and HP in particular, started looking elsewhere. HP appeared to loose interest in the UNIX market place, allowing both their own products and the subsumed products from DEC/Compaq (and to a lesser extent, Tandem) to fall into the legacy category. They produced Itanium based servers, but they were never up there with IBM, except in the very-large system market. Only customer pressure has kept many of the OS's alive.
In the meantime, IBM has been left with the only non-Intel/AMD UNIX offering that was actively being developed, and as a result, has kept market share. Even though there has been no real competitive pressure, IBM has used the convergence of the AS/400 and RS/6000 lines, and to a lesser but significant extent the z series, to move the architecture forward. They have borrowed from other IBM systems (and their competitors) to introduce type 1 hypervisors, hosted application partitions, and a pretty much unrivaled virtualization capabilities. The supported filesystems have scaled, the support for other technologies such as SAN and SVN has gone hand-in-hand with other IBM products.
AC - thanks for reminding me what a disaster Rick Belluzzo was for the computer industry. And don't forget, his reward for running SGI into the ground was a job at ......... Microsoft.