Q&A: Dell eyes Linux future despite desktop retreat

ROUND ROCK, TEXAS – When IDG News Service last week sat down with Dell Chairman and CEO Michael Dell, it seemed appropriate to ask him about some of the not-so-mature markets Dell has pursued. In the case of the Linux desktop, Dell pulled back on large investments in 2001 as demand faltered, Dell said. But this failure has not stopped the company from attacking the nascent markets for high-performance computing clusters (HPCC) and for Linux clusters running Oracle database software. Dell painted a picture of where and when he draws the line on new ventures.

Dell traditionally goes where the money is and toward established markets. You are currently going after things like Linux clusters running Oracle and high-performance computing pretty heavily, which seem to be more of niche areas right now. When do you decide to enter these types of businesses?

Customers decide more than we do. If enough of them want it, that really motivates us to do things.

We have actually been playing around with this clustering market for quite a number of years with a bunch of PhDs we have. It’s a powerful idea for us because essentially what you are doing is taking industry-standard building blocks, putting them together and building a supercomputer.

We believe we already have leading share in the HPCC market. While it is a pretty small market today, it is growing very rapidly.

I’ve heard your stable of PhDs mentioned quite a few times in interviews. Is high-performance computing an area where you are putting some heavy research and development funding?

Dell has over 3,000 people developing products. We spend half a billion dollars on R&D. So there is a fair bit to what goes on here in developing products. Some have suggested that Dell does not do any R&D, but that is not the case. We have patents, we have buildings full of people that do this stuff. You can do a lot of with half a billion dollars of R&D.

More importantly, we probably, more than most, embrace this idea of collaborative development. You have partners like an Intel or a Broadcom that has an ingredient technology, and we leverage those partners to deliver a solution to our customers.

How does that notion of collaborative development play out as Intel servers bulk up and become better systems? HP and IBM typically do more work on tuning their Intel systems with in-house technology. Can you continue to compete as effectively without spending more on research?

Well, we will do as much as we need to do. It’s as simple as that.

If you dial the clock back five years ago, you could have asked the same thing. You can always ask that question. The issue is that, in our industry, technologies always evolve to standards, and when that happens markets become larger, products go to volume and customers benefit. There may be some companies out there that don’t want that to happen, but we are not one of them.

So you don’t see an eventual need to distinguish your systems from those of competitors that are putting their high-end Unix knowledge in Intel systems?

You look at the troubles that some of those companies are having, and I think it directly relates to the fact that people are not willing to pay for the uniqueness. There is this notion in the industry that if you spend on R&D it is a good thing. Well, that is true, but it may be a simplistic way to look at it. Let’s say you spend 100% of your revenue on R&D. Is that a good thing? No. So, too much is not a good thing.

There are two kinds of R&D, or two kinds of innovation. There is innovation that benefits the customer and then R&D that benefits the supplier and locks customers in.

Would you consider things like N1 from Sun and UDC (Utility Data Center) from HP examples of these vendors doing research to try and keep customers on their platforms?

That is where I think a lot of companies have gotten themselves into trouble. They have tried to develop proprietary hooks that do not benefit the customer. The customer is basically saying, “I don’t want it, and I am not going to pay for it.”

While their visions are far off, Sun, HP and IBM do seem to want to work on managing multivendor networks and not just their own platforms, but you don’t seem to buy into that at all. I have not heard anyone from Dell articulate this type of management platform.

In terms of systems management, there is a lot we have done so customers can manage our products and other products and link into system management platforms like HP’s OpenView. There is no one right answer for all customers. We are all working to have better systems management.

In terms of utility computing, I think it is one of these amorphous terms that means a lot of things to a lot of people. We have a managed services business that is growing very quickly and sometimes people think about that as utility computing.

On the storage side, you have partnered with EMC to offer your customers a full spectrum of storage from the low to the high end. Why not create a deal with a Unix server vendor as well, to fill out your product line and perhaps open up new services opportunities?

Who do you have in mind?

A company like Sun or IBM sells lots of boxes onto EMC.

We are selling some EMC storage that attaches onto a Sun box, but mostly where we see Sun is with Linux replacing Sun. In that sense they are more of a competitor.

Obviously, the Unix guys are competing with you, but Dell had been competing with EMC on storage as well. When do you decide the time is right to partner and when the time is right to compete?

First of all, we are focused on things that are broadly accepted standards. I see proprietary Unix going away and being replaced by Linux and Windows. That is kind of what the market trend really is. I grant you there are customers still buying those things, but I think they continue to get pushed up to the higher end, and the volume Unix market will be predominantly Linux on high-volume servers like those from Dell.

When do you see 64-bit computing being a reality for Dell then?

Of course, it all depends on applications and the cost of the processor. Right now the number of applications available for 64-bit is kind of emerging.

We will be in that market. It will start out as a high-end market. As there is more volume, it could get down into the more mainstream types of applications.

Have you made a commitment to a company like OEM Newisys to ship servers based on AMD’s 64-bit Opteron chip?

No. Our strategy has not really changed. We could have 64-bits in both servers and desktops. It’s clearly more likely to occur first on the servers. I don’t hear a lot of people asking for 64-bits on the PC.

Looking back, you were one of the first companies to have Linux from the desktop to your highest-end server. You also made a lot of investments in companies like Eazel that were working on making desktop versions of Linux easier to use.

Some of it worked, some of it didn’t. We did make a lot of progress, but unfortunately the desktop Linux market didn’t develop in volume. It’s more of a server opportunity.

Some people at Eazel (now defunct) said that Dell was still willing to invest in the company.

Yeah.

Doesn’t your commitment to Eazel there seem to contradict your decision to pull Linux as an option for desktops in 2001?

Well, I think there might have been a misunderstanding of what we pulled and didn’t pull. If you think about the breadth of our product line, you have a 150 different models. The demand for Linux on the desktop is there but it’s not huge. So to support Linux on the desktop for every single model does not make economic sense. So what we have done, effectively, is offer it on a smaller set of configurations.

I was under the impression that customers had to request Linux for desktops in large quantities as a special order.

I think on our workstation products you can buy it off the Web site. We found that (the demand is) in the education and scientific markets… and they want it on workstations. But again, we sell systems with free DOS, so you can put your own OS on there.

I am sure you saw Dell mentioned in the Microsoft antitrust case where a Microsoft memo spelled how upset they were at your decision to sell Linux on the desktop.

Whoo! (Twirls finger.)

A large chunk of your staff that dealt with Linux was fired after you dropped it from the desktop and some people suggested Microsoft may have nudged you in that direction. But you are saying there is nothing to that speculation?

No, we offer Linux on the desktop. We continue to offer Linux on the desktop and there is nothing else to say. Microsoft is a great partner of ours, and they don’t sell only to Dell, and we don’t sell only their products. It’s pretty simple. (Laughs.)

It does seem like Linux has come a long way on the desktop this year. Do you think it can be revitalized there?

I think you can look at our prior activities here and know that this is something we are interested in seeing, but if you asked if it would be significant in the next few months, it doesn’t appear that way. We are not opposed to it. If there is a volume Linux-on-the-desktop market, then great.

Dell was one of the companies that was pretty vocal about being happy that HP and Compaq were merging. Now that the chaos of the merger has passed, what are the effects you are seeing?

Last quarter our business grew 28% in units, and the rest of the industry grew 2%. I think the company you mentioned had negative growth in units. So, I think we did pretty well.

Do you think IBM and HP have the right model as the one-stop shop for everything? Is that model, along with heavy services, the one that will survive?

I think our results have been pretty good this last year. I think our results speak for themselves. We’ll leave it at that.