Click here for HP sponsor message

Cautious hope surrounds IA-64 news

All applaud new direction, but customers, vendors look for software to seal value

Details on the migration and the designs from CSY R&D chief Winston Prather
Editorial: A 64-bit tonic to turn back the years

In companies not far from the HP 3000 headquarters, MIS managers and channel partners were glad to hear of IA-64 futures. But all said that announcing new hardware and operating system plans demanded concurrent IA-64 commitments from HP software, as well as from third party suppliers.

“It’s great news for application providers who have HP 3000s to sell,” said John Burke, systems manager at Sacramento-based Pacific Coast Building Products. “But it has to tempered by the realization that it’s software and applications that sell systems.”

“There’s got to be more. This kind of commitment is what’s been missing for the last half-dozen years, a feeling that HP is committed to the platform for the long term. Now there’s got to be a real push from HP to get developers to write for the platform. If that involves them going out and identifying niche markets where somebody has software they need help promoting, then they need to do that.”

Burke doesn’t doubt for a minute that HP will be able to migrate its user base. “They do have a track record in going from CISC to RISC,” he said. “They more than anybody can pull it off, because they’ve done it, and in fairly dramatic fashion. In this timeframe, I don’t think they’re going to get caught in the performance crunch they got caught in last time. I think there’s enough headroom in PA-RISC.”

“I think some people switched sooner than they probably would have liked last time, because they didn’t have any more room to put Series 70s in, running out of physical floor space. So some of the bigger shops suffered a bit.”

“This is great news, but in the absence of new applications, we’re not really much further along than we were two weeks ago,” Burke said. “The future of the 3000 doesn’t look much different.”

Donna Garverick, senior systems programmer at Long’s Drug Stores and chair of the SIGSYSMAN special interest group, said she was excited at the prospect of being involved in the upcoming migration, but watching to see how software providers would set costs.

“All in all, I think this is really good news,” said Garverick, who leads a group of systems managers in the Interex user group. “It’s nice to hear that HP’s going to take a big step like this, because it shows HP’s commitment toward the 3000. But cost is going to be a big issue. I’m concerned about the costs of moving to the new systems.”

Those costs include fees to upgrade software, she noted. “When we did this last time, it wasn’t so much the costs of bringing in the new hardware, it was the licensing fees on the software – and certain 4GL providers, made it so some people just couldn’t move forward.”

“We may not be seeing the huge price jump we did with Spectrum chips, because this is a joint venture between HP and Intel,” she added. “I’m hoping this resurgence of the 3000 isn’t coming with a cost, like ‘we love you if you’re an IA-64 customer and we don’t otherwise.’ I don’t think we’ll be getting that out of HP.”

Garverick works with the retailer’s remote support teams to administer almost 400 HP 3000s, including hundreds running on MPE V. “It’s going to be a challenge and half for us,” she said of the proposed migration. “I don’t know if the MPE V is going to run on the new chip. I suspect I know HP’s answer, but for someone like Long’s it would be an important answer to know.”

“I think we would wind up migrating our 995/300,” she said. “We need to decide exactly how we’re going to do that. It’s a good move [to provide a new box that accepts both PA-RISC and IA-64], if nothing else because it spreads the cost around.”

A lack of schedule yet for the release fits well with the 3000 customer, Garverick added. “HP may know us better than we think. Give us three years to get over the shock, and then we’ll be ready for it.”

While noting that HP has said no major subsystems would be left behind, Garverick added, “I can hear the voices from IPROF saying, ‘What about HP C?’ It will be interesting to hear HP’s answer on that.”

“I’m kind of waiting to see how it hits the fan that there will be two versions of MPE/iX,” she said, “and that the PA-RISC version would be phased out.” HP has supported the MPE V operating system for 11 years after the PA-RISC hardware shipped, “and I hope they’re willing to do the same thing for the PA-RISC version of MPE/iX,” she said. HP traditionally has such a good attitude toward supporting the customer base on the 3000 side.”

“For the really big customers, this is exactly what they need to have. HP has got to be able to position the 3000 to be able to churn stuff through fast enough to keep the big customers happy. The little customers will benefit because there’s a more powerful architecture out there. I just hope price isn’t a factor that keeps them from enjoying the same benefits.”

Analyst okays

The Aberdeen Group’s John Logan predicted in early 1996 that HP would decide to use the new architecture in HP 3000s. Analyst Bob Sakakeeny of the Aberdeen Group was just as interested in the news about the sales growth of the 3000 – which prompted the IA-64 decision.

“It’s intriguing that the so-called older platforms like the HP 3000 and the AS/400 are getting some new customers,” he said. “It makes sense to upgrade the 3000 because of that. They’re reassuring their installed base that the product’s going to be around for awhile, that it’s not going to end when Merced comes out.”

HP’s return on this investment is being calculated differently this time. “It’s what I classify as opportunity costs, what you lose if you don’t do it,” Sakakeeny said. “If you don’t reassure the installed base there’s a growth path, they’re going to jump. When you show them there’s a way up, you stop any kind of erosion of the installed base.”

Sticking with the HP 3000 is a safety issue for customers, he added. “The folks who are relying on products like this to run their business are much more interested in stability right now,” he said. Expecting the new environment to bring new applications may take some time.

“The application question for IA-64 is open, irrespective of the operating system,” he said. “It’s going to take a long time for 64-bit apps to take advantage of the hardware. You’ll get some performance increases by moving a 32-bit app over. There’s not a huge demand for 64-bit apps right now.”

Delays in delivering Merced chips won’t have much impact for the moment. “This is a humorous guess, but I’d assume the folks in Redmond are asking Intel to slow down as well, to wait for 64-bit NT,” Sakakeeny said.

Analyst Bill Moran of D.H. Brown said HP has done “an outstanding job” in nursing along the 3000, “and this is a good example. This ought to make their customers reasonably happy about what’s going on. If the 9000 moves to IA-64 and they didn’t do that, there would be a lot of questions. When the 9000 moves off PA-RISC, they’re going to pull the plug on PA-RISC, so you’d have to question what the 3000’s future is if they didn’t do this.”

“They’ve been keeping the 3000 relatively current. This is another step, albeit a bigger one, in the same direction. It sounds like the user base is growing again, and shipments of this system are rebounding.”

“The advantage the 3000 has is that it has been out there a pretty long time, and it’s a pretty mature thing. If you were to attack certain areas, you’re comfortable it’s not slideware. You can go to Microsoft and get a slideware pitch, and with a lot of vendors you’re not really clear what’s shipping today. I don’t think you have those questions about this platform. If you’re looking for a tried and true solution, I think it brings a lot to the table.”

It appears you will not be able to migrate from the most current version. At any moment of time there’s a current version of PA-RISC. It doesn’t appear to me, my best guess is you will not be able to migrate from the most current RISC platforms to IA-64 with Merced. That isn’t going to fly. From a logical point of view you could do it, but from a performance point of view it seems that’s not going to be quite what you want. It’s not going to be blazing away and dominating the best RISC at the time. Intel led people to believe that it would, if you listened to their hype along the way. The reality is that it isn’t going to happen, and there’s another chip coming. If you’re pretty current with PA RISC, you may not be able to do an easy flip over to Merced. If you’re not exactly current, say a PA-8000 based or PA-7200 based, then they might be able to do the Merced segue. If you’re really hungry for performance, you’ll have to stay with PA-RISC for awhile until IA-64 is a clear winner. They’re now saying that doesn’t happen with Merced, maybe it happens with McKinley. If HP is clever, they can manage the delay of Merced.”

“They moved the HP 3000 to PA-RISC, and they have some experience in doing this. The other thing that 3000 guys are going to get is letting the 9000 customer being the trailblazer. They’re going to debug it on the 9000. They’re telling the customer they’ll come along once all that is ironed out, they won’t rush into this. I think that will be well appreciated by the customers. A customer on the HP 3000 by definition is not an extremely adventurous customer. If they were adventurous, they would have gone and done something else.”

When you tell the 3000 customer you’re going to do this in a way that minimizes the risk, and on a strategic product like that, that’s all goodness. I don’t see much downside there for HP.

Moran thinks the sales boost for 3000s as a result of the announcement is “marginal, but I think they’re done a decent job on the 3000. My take is they don’t have a lot of unhappy customers out there – because they took it to PA-RISC, and rationalized the pricing so the customer didn’t feel like they were robbing him because it was proprietary. Showing the HP 3000’s going to live into the next century will probably get them some business.”

“If HP can’t migrate their leading edge customers to Merced, I can live with that. But if the next IA-64 chip doesn’t do it, I think they’ve got a big problem. Then you’re raising a lot of questions about the overall strategy.”

ISVs on board, watching

Duane Percox, founding partner of K-12 school application supplier QSS, said the HP move was a welcome one, and said 3000 customers may benefit from following HP-UX users of the architecture.

“I’m excited that HP is moving in this direction,” Percox said. “It shows a commitment to the platform and standard for hardware. HP’s decisions have shown that it needs to move in a commodity model for hardware, so the real value appears in the operating system.”

“It’s good that first generation of chips won’t be on the 3000,” Percox added. “Those first generations won’t be all that fast anyway. CSY has always been conservative in these things. The PA-RISC migration was the one they bet the bank on.”

Lee Courtney views the IA-64 commitment from a unique perspective, as HP’s former MPE Product Manager during CSY’s last migration on the Spectrum project.More than a decade later Courtney heads up Monterrey Software, a 3000 security solution provider.

“I think it’s pretty exciting,” Courtney said. “The 3000 folks seem to be coming into the Merced thing rather late, but that might be a good thing. For one, the 9000 people are going to shake things out. PA-RISC is going to have a lot of life left in it, even on the high end. I wouldn’t be surprised if HP continued to maintain some imvestment in PA-RISC just to hedge their bets on Merced.”

The software challenges are significant for the division to succeed in the project, Courtney added. “CSY has lost a lot of talent, kind of like the 3000 has been allowed to wither on the vine. A lot of expertise has left that organization that they’re going to have to build. The most glaring example is languages, a place they’ve fallen way behind on. For Merced to be successful, they have to give us the tools to take advantage of it.”

“I certainly think it’s going to be a lot brighter,” Courtney said of the 3000s future. “I think the 3000’s going to be around a lot longer anyway, based on what people do with it. With Merced, we might even see some more new applications. This announcement adds a shot in the arm to the product line. You may not see any immediate benefits from it, but in the long run it will be good for the platform.”

"I think moving MPE to Merced will be comparable to what IBM did with the AS/400 when they went to PowerPC,” Courtney said. “It will be interesting to see if they do anything with it at the low end.”

“There’s always been constant pressure, even when I was with HP, at the the high end. But it will be interesting what they have to say about the low end. If they don’t say anything, fine. But they’re going to have to do something about the low end eventually.”

“When we were at this point with PA-RISC and we announced it, there were literally hundreds of people working on the migration of the 3000. There was a migration plan, how we plan on getting you there. None of that stuff is here right now, and I wonder what we read into that.”

“Anyone who was marginally attached to the 3000 has already left. If there was anyone who wanted to leave and needed an excuse, they have already left. Folks who don’t really care what platform they’re running on, except it gets the job done, those are the ones who will get a lot out of the announcement.”

Courtney said that supporting his software for the new architecture “is a no-brainer. That’s gonna happen because that’s where the high end 3000 datacenters are going to be, and that’s who our primary market is. HP is going to have their hands full with the OS, subsystems, languages and I/O tested put together. I think they’re going to have to leverage a lot of third party solutions. Compared to what went on with the migration from the stack architecture to PA-RISC, they just don’t have the brainpower in place. They have people who have worked on the 3000 a very long time, but they don’t have as many of them.”

Experience with moving a customer base is the best advantage HP 3000 customers are going to enjoy, he added. “The nice thing that the 3000 people are going to have over the Unix people, and it’s going to be really apparent, is the migration story for upward compatibility,” Courtney said. “I expect that CSY will have the same migration story that they did from stack architecture to PA RISC. People went through a lot of pain going to HP-UX 10.0, and I expect they’ll go through at least the same amount of pain going to Merced. The 3000 people will at least be able to say ‘Thank God we aren’t there.’ It’s going to validate the decision a lot of people made to stay on the [3000] platform.”