Course corrections at IBM server group
Bill Zeitler has to roll up his sleeves and fix Big Blue's server product strategy--in a hurry.
- Shankland covered the tech industry for more than 25 years and was a science writer for five years before that. He has deep expertise in microprocessors, digital photography, computer hardware and software, internet standards, web technology, and more.
In nearly six years as leading IBM's systems and technology group, Zeitler helped restore Big Blue's crucial mainframe business, embraced Intel chips and Linux as competitive weapons, launched the industry's top blade server design, and led its Unix systems from a distant third place to the top spot in 2005.
But the first quarter of 2006 wasn't so hot for IBM's servers. Mainframe revenue shrank 2 percent and Unix server revenue declined 6 percent, according to Sanford C. Bernstein analyst Toni Sacconaghi. IBM's Unix business slumped at the same time that previously down-and-out rival Sun Microsystems couldn't keep up with demand for its newer models. Not only that, but IBM delayed its upcoming Power6 processor into 2007.
Zeitler, 58, is preparing his counterattack, though. IBM will cut Unix server prices Tuesday, he said. IBM on Thursday launched a new midrange mainframe in China in an attempt to hook a whole new market on the powerful but pricey machines. And in the longer term, Zeitler sees strong potential for servers using Advanced Micro Devices' Opteron processor.
Mainframes--systems with powerful communications ability, flexibility and reliability--had been given up as a relic from an earlier age, but IBM has rejuvenated the line. Its new mainframe, called the System z9 Business Class (BC), is the cheapest mainframe so far. In May it will be available with the z9 Integrated Information Processor, or zIIP, which lets customers run IBM database software without incurring per-processor charges for other software.
To reinforce its mainframe modernization and market expansion, Big Blue announced a new Shanghai mainframe development lab working on Linux and other technologies. Zeitler talked about the venerable servers and other priorities in a meeting with CNET News.com's Stephen Shankland.
Q: Could you talk about the new midrange mainframes coming out?
Zeitler: We're announcing the new system z9 business class mainframe. It brings all of the security, reliability and scale of our traditional mainframe products to a new class of small and medium enterprises at the lowest entry point we've offered.
Q: What is that entry point?
Zeitler: Well, it's about $100,000. That's considerably below where you would have been able to buy a mainframe traditionally.
Q: I suspect they probably aren't too much less expensive to manufacture. So what's the rationale for lowering the price?
Zeitler: In fact, the Business Class processors are, from a cost standpoint, better positioned than the traditional z9 processors. We want to be able to offer both midmarket customers and customers in the emerging markets, who are building out their infrastructures, the same kind of reliability and security that traditional buyers have been able to get--without having to pay an enormous financial penalty to get into this product.
Zeitler: One thing that happened is we didn't close as many large transactions as we had anticipated. We had some particular weakness in the Americas. Beyond that, though, I think that the important point is that we've grown revenue three of the last five years...I expect with the product offerings and some adjustments we've made in our sales approach, we should be able to get this thing back on track going forward.
Q: The other soft part of the quarter was pSeries--the Unix market. You managed to get the top spot in the Unix server market in 2005 in terms of revenue. What happened, and what are you going to do to fix that?
Zeitler: I think you could fairly well say we shot ourselves in the foot with pSeries in the first quarter. Much of it was around the introduction of our new Power5+ (processor) family of products. Any time you introduce products at the low end that have very strong capabilities, you have some susceptibility to cannibalizing products above them, and that's what happened...We have made some changes relative to that...and I think we'll be in good shape.
I don't have any doubt that the new (mid-range Unix servers) are extraordinarily strong technically. We just didn't do the job we needed from marketing and sales positioning.
Q: What's happening next Tuesday?Zeitler: We're going to announce some pricing adjustments, some ISV (independent software vendor) incentives and some other positioning things to make these pretty attractive offers--particularly to ISVs and to business partners...who help us to take these product to the market.
Q: A couple of years ago, IBM said it expected the Power6 processor to show up this year. Now it's a 2007 phenomenon. I presume that the Power5+ chips will be moving to the high end of the Unix server line sometime later this year. Why did that scheduled slip come about?
Zeitler: I think that our CFO, when he (announced) the first-quarter results, said that we expected to have the Power5 plus transition done in the third quarter across the whole line. We have moved the next-generation products into the beginning of next year. Candidly, that gives us time to make sure that we get everything done in the way we need to. The chips coming off of the 65-nanometer (manufacturing process) line are doing wonderfully well, but there's a lot more to this than chips.
The other point of it is that the first quarter is a better time to introduce products like this than the fourth. The Power5 and Power5+ families still offer by far the best benchmarks. This is a business judgment, and I think it will prove to be an OK thing to do.
Q: Sun has been pretty weak and losing share in the Unix market for a quite a long time now. For once, they actually have pretty strong demand for their new UltraSparc 4+ systems. Are you seeing them as more competitive now, or did they just get lucky with the product cycle?
Zeitler: I see more interest in their Opteron-based products. I think that those products offer better competitive choices than at the high end. I don't see them as being of particular competitive challenge, compared to what they have been. But I don't think there's any question that they have adopted Opteron, they have a stronger offer in the market than they did with their traditional offerings.
Q: And do you see their UltraSparc T1 "Niagara"-based servers, the T2000 and T1000, in the market at all? Those are pretty new and kind of freaky.
Zeitler: Well, at least from my standpoint, we don't see them in competitive situations terribly often. I personally think they'll be a little bit more of a niche product and not something on which I would expect to see them gain a lot of market share.
Q: How about the other side of the competition? Hewlett-Packard says Itanium is back. Larry Ellison says Itanium with HP is the most important platform for Oracle. Do you see Itanium systems from HP more often?
I don't see anything that could convince me this product is any stronger today than it was six months ago or six years ago. I don't think you see very much of their extension outside of HP's base attacking products that we offer. You have to take these things with some healthy skepticism.
Q: You mentioned that the Sun Opteron systems have given them more of a presence in the market. You have Opteron systems--you were the first tier-one manufacturer out of the gate--but you sell them for the high-performance computing niche as opposed to the mainstream server market. What would lead you to change that position?
Zeitler: The 64-bit extension offerings from AMD and from Intel (have shown an) unbelievably high rate of take up. That's an indication that what the market was looking for is not another 64-bit architecture like (Itanium), but a rational way to move the x86 architecture from 32 bits forward to 64. It's what makes AMD and the Opteron activities that Sun and HP have been pursuing so important. In that light, we have been looking very, very carefully at what HP has been doing and what Sun's been doing.
When we started this, we got a lot of interest from Wall Street clients in particular on using Opteron products for high-performance computing, and so we brought out one new product do that. We followed it with a blade because there's a lot of market there. We've seen really good take up on the Opteron products we've had. What would move us to a (higher-end model) is the continued acceptance of those products from our competitors, and honestly, they are having pretty good success.
If you look at HP's results over the last few quarters, it wasn't their unit growth that was causing them to improve, it was their average unit revenue, and the average unit was improving because they had more AMD content than they had had previously. Even in our own case, the average unit revenues on an AMD blade are much higher than the same kind of Intel blade because the performance is better, and because the performance is better, people put more I/O (input-output components such as network cards) and more memory and other things on them. There are a whole bunch of business considerations that would say we would have done better had we had a four-way Opteron product for sure.
Q: Does that indicate that you are going to head in that direction?
Zeitler: I don't want to speculate when and if we would expand our activities here. But there's no question HP and Sun have benefited by having a broader Opteron portfolio than we've had.
