Jul 29, 2003 (08:07 PM EDT)
The Farmer In Dell
Read the Original Article at InformationWeek
Dell Inc. is closing in on its 20th anniversary as an influential provider of information technology. CEO Michael Dell might not have been able to foresee this level of success two decades ago, but he feels he's got his finger on his customers' pulse. InformationWeek editors Paul Travis and Larry Greenemeier recently sat down with Dell to talk about a possible economic rebound, "scale-out" server architectures, and close contact with customers.
InformationWeek: How do you see the economy and IT spending? Is it coming back to life?
Michael Dell: If you look at the last three quarters, it's been slightly positive, certainly better than it was if you go back four or five quarters ago. We've been in the fortunate position where we've been growing even while the market has been down. If you take Dell out of the market, it goes down. During the first quarter, we grew units in 29%; the rest of the industry was down 1%. It looks like there are incremental signs of improvement here and there, but not a massive rebound or recovery.
InformationWeek: One of the things we've been hearing from a number of IT managers is their desire to consolidate a large number of servers to a much smaller number. What's the challenge in terms of making your scaling-out approach more appealing than a competitor's approach of selling a scaled-up, Intel-based server with lots of processors?
Michael Dell: I think the problem you're highlighting is that the original idea of client-server computing was, "All right, get an application, get a server." The problem with that is that all of a sudden you end up with 3,000 servers and you've got one per application. This one is used 5%, this one is used 20%, this is used 30%, and this one is used 0%. That's not what we're talking about. What we're talking about is basically aggregating all the computing power and being able to dynamically allocate resources among a pool of shared servers, as in a cluster. Most likely, they're all in the same data center physically and attached using very high-speed, serial-type interfaces.
So scale out is really leveraging the high-volume industry standard economics of microprocessor chip sets, high-volume disk drives and other components built-in hundreds of millions of units per year to run large databases and applications. Oracle and SAP are really embracing this scale-out idea not only for better price/performance but also for better availability and better scalability.
InformationWeek: Are you seeing a lot of demand for eight-way or 12-way Intel-based servers?
Michael Dell: If you look at the demand for eight-way servers, it's gone way down. The industry forecast for eight-way servers has gone way, way down partly because four-way servers can now do the job of eight-way servers. This is one of the problems with scale up. Let's say I'm a really smart engineer and I've developed this fantastic 64-way server. I've got to develop this microarchitecture and memory and chipset; it takes years to do and by the time I got this thing--oh, there's another processor that's four times faster, so my 64-way is now just as fast as this other guy's 16-way.
InformationWeek: Are the management tools available today to make scale-out computing a reality for businesses? Or is this an area that needs to be a lot of work?
Michael Dell: There's work to be done in systems management. There's work to be done in aggregating all parts of the network infrastructure from storage to servers to networking itself to application management. Really, no company anywhere today can fulfill the grand visions that are out there either for scale up or for scale out. We're all continually working to fill the opportunities that are out there. But it's really a question of, do you believe in 64-way servers aggregating all the power, vs. the high-volume industry standard servers that continue to take on more and more computing tasks. Probably the best indicator of whether something is working or not is do people actually buy it? And people are buying our stuff. They're buying it in huge volumes and our server sales continue to grow at a tremendous rate, faster than anyone in the industry.
InformationWeek: Dell has gotten into a lot of new areas over the past few years--storage, networking, printers, and PDAs. How has that worked out for the company, and do you see any areas that might be ripe for Dell?
Michael Dell: Certainly if we look at our business, the fastest-growing areas have been the enterprise servers, storage, and services, not to mention some of the new software peripheral areas--networking, printers, PDAs, and projectors. As we look across the $800 billion IT market, there are lots of opportunities. But we have a lot of on our plate right now, so I don't know that you'll see a massive number of new things. I think the vast majority of our growth in the next couple of years will come from things that we already have. Printers, I think, will be a very significant business for us in several years. Right now it's not. But it's growing very quickly; it's well above our plan.
InformationWeek: Regarding the Linux market. You've seen some success selling Linux servers in the high-performance computing market. Do you think The SCO Group's lawsuit against IBM for allegedly misusing its Unix license to improve Linux will create any sort of chilling effect for companies unsure about the fate of Linux?
Michael Dell: There are some companies I would say that are sort of hypersensitive companies who will ask about it certainly, but not a significant number.
InformationWeek: Are you seeing much demand for Linux on the desktop?
Michael Dell: We see some demand for Linux on the desktop. It's traditionally been in the workstation or in the university world. It hasn't been broad-based.
InformationWeek: How important is customer input when you're developing product, and how do customers provide input?
Michael Dell: If you look at our product road maps, basically those are validated by customer-advocacy sessions. Our engineering teams come in and say, 'So here's what we're thinking about for next year for storage, for desktops, for workstations, notebooks. Here's the plan, what do you think?' There are listening posts all over the company; we don't do things without the customer input. We also have a tremendous amount of data. Remember, we're selling 25-26 million computers a year, so you have tremendous customer information that is really unfiltered. With a channel mode, I say, 'OK, you guys go sell all this stuff.' Well, the customer comes to the channel or the channel goes to the customer, but do we really know all the things that went on. If something wasn't quite right, do we really know? No, we don't. We just keep building the stuff. And so you're just completely disconnected. That is the worst nightmare for us. So we want to be in one-to-one contact with every customer we have: little customers, big customers.
InformationWeek: What's the greatest advantage of these customer-advocacy sessions?
Michael Dell: First of all, I think they really appreciate the opportunity to provide input. The second thing is that by getting their input, you have a much higher return on your research and development. If you look at the ratio of Dell's R&D to profit, it's five times that of our major competitors. Why? Because we know what to develop. If you develop something at Dell that saves the customer money, you're a hero. On the other hand, if you develop something at our competitors that's really cool, but nobody wants to buy it, you're still a hero. Here, if you develop something really cool, but nobody wants to buy it, you screwed up. Who did you talk to? Where was the input? What did the customers tell you? I'm not saying it never happens, but the accuracy rate on our methods is very, very high because the customers are very involved. The customers know our products because they were there telling us to develop them, saying, "No, we don't want one power supply in this server, we want two power supplies. We don't want this much memory; we want that much memory." They give us the input and we go do it.
We also find that there are customers who are always kind of pushing the leading edge of usage and capability and challenging us. Those are the best ones because we can learn from them. In certain markets, like Japan, they tend to lead into new features or client systems. And first you can look at that and say, "Well they have different requirements." Well, they aren't different; they're just ahead of the rest of the world.
InformationWeek: You're coming up on the 20th anniversary of the company. You've seen a lot of changes, a lot of things come and go. Do you see anything in particular that will have a high impact over the next several years? Maybe wireless changing the way people do computing or a combination cell phones and PDAs changing the way people communicate?
Michael Dell: I think there's a couple of ways to think about that question. If you think about the period that we went through in the 90s, the sheer growth and development in the industry was so massive that you had a lot of these things that sort of fit into a characterization or stereotype that you're talking about. The industry is now much, much larger, so to move the needle to that magnitude is a much more difficult thing to do. So, for example, a shift to wireless is massive and very, very important, but if you asked how many more units it represents, that sort of all gets gobbled up in the enormity of the industry. The kind of embryonic period that a lot of us went through in the 80s and early 90s--that was sort of a lost era; the industry only gets started one time. There will be new industries that are started within IT computing. But that was sort of a special thing.
Photo of Michael Dell Courtesy of Dell Inc.