Steve Mills, senior vice president and group executive for IBM"s software division, discussed Linux usage trends, the company"s involvement in the development of Linux and open-source virtualization software, and plans for AIX during an interview with Computerworld in San Francisco Wednesday at the LinuxWorld Conference & Expo.

Part 1 of the interview follows. Part 2 will be available online Friday.

To what degree are you starting to see fresh deployments of Linux, as opposed to customers who are merely switching from Unix to Linux? They"re still doing what they were doing, but they"re also now doing a lot of incremental new implementations. What has been building momentum in the marketplace for the last couple years has been actually new applications going in that are running on Linux. It"s not easy to track what a lot of independent software vendors are doing. But there are many, many hundreds of applications out there today. These application providers are putting their code onto Linux because they want new sales.

The way you want to think about this is the way the evolution of Unix occurred. That was a very long process. The early days of Unix tended to be very simple applications, a lot of one-off things that people were cobbling together. There was a certain freedom of expression if you will associated with some of that early application activity on Unix, and then it began to come together around more of the classical application function that businesses need to run. As you go into companies of various kinds today, in the Wall Street world, you"ll find all kinds of analysis and trading system activity that takes place on Unix systems. If you go into the telco environment, you find a lot of systems there associated with the whole front operation, a customer-facing environment, billing capture off the electronic switches and things of this nature -- production control, plant-floor systems, all kinds of mainstream uses of computing that are based upon Unix.

The Linux movement bears resemblance to this. If you go back to the very beginning, in the early "90s, you had a lot of the one-off things taking place off onto the side. The easiest thing to do was to take some existing work of a lightweight nature that wasn"t going to stress Linux very much and move that over. Linux is less expensive. I can run it on a cheaper box. I can save some money. So it"s targeted at some reduction in ongoing expense from simple applications.

And then we transitioned into more sophisticated environments, where people are running their SAP ERP environment on Linux. They"re running their call center on Linux. They"re running their retail store operations on Linux. It"s moved from that early adopter stage into more of the mainstream now.

Now that Linux is maturing, are you seeing more companies trying to virtualize their environments? There is clearly a trend toward trying to get better cost of ownership, right. The use of VMware or the Xen open-source hypervisor, these kinds of things are building up momentum in the marketplace because the customers are staring at cost of ownership and often relatively lightly loaded individual servers and saying, "Is there something I can do, some additional code I can insert that"ll help me get more usefulness out of that particular server?"

There are many things in this category. We sell our BladeCenter offering into this set of requirements. We sell our pSeries with AIX and partitioning into this set of requirements. And, of course, we support VMware on our hardware as well. And there are customers that choose it as a way to get more leverage on their server spend. It"s not a Linux-unique phenomenon. I don"t see any less popularity around this idea for Linux as anything else.

To what degree is IBM working on Xen? We have internal activity related to Xen. We are engaged with the community on Xen. We"re anxious to see Xen become more mature. We think the market needs choices on hypervisor virtual machine capability. An open hypervisor certainly has a lot of attractiveness.

Will Xen be IBM"s recommendation for customers, once it"s fully cooked and ready? We wouldn"t be engaged in it if we didn"t feel that we were going to eventually be putting a big endorsement around it. In a sense, through our involvement, we"re endorsing it today. We"re not telling people not to start projects with it, not to experiment with it. But you"re right. It"s still going through its earlier stages here of building some maturity. But I think the pathway towards a more mature and sophisticated Xen code base is one that you can plot out.

To what degree is IBM involved in the development of the Linux kernel? We have our Linux Technology Center, and we have a team of people that are involved with the Linux community on Linux development and have ... since "98 or "99. We don"t have any people that are literally part of the core team. We play an influencer role in terms of what we work on, suggestions that we make, code samples that we offer up. We"re trying to encourage a particular set of improvements to Linux. Obviously, our interests tend to relate to things that our large corporate customers are most interested in. They worry about robustness, scalability, fail-over, so that"s a direction that we have an interest in. Different groups sort of stake out areas that they have greater interest in. We"ve had very good success in having ideas that we"ve been working on become accepted by the community and incorporated into the kernel.

How do you feel, going forward, about Linux being a two-horse race between Red Hat and Novell? There were always smaller distributors, but this two-horse race phenomenon has been with us now for some time. We"re perfectly fine with it. It"s about distribution. It"s about accessibility by the buyers to a fully packaged operating system. The ecosystem works. One thing we worry a lot about is we don"t want to do things that disrupt that ecosystem. This market has good momentum, and the role that IBM plays in both enabling our technology to Linux and doing a lot of things to help customers and encourage Linux adoption plays very well to our needs and requirements. We don"t need to be a distributor to accomplish those things. The Linux market is many, many times bigger than Linux itself. There"s more money surrounding Linux than there is in being a distributor of Linux operating systems.

Given the growing popularity of Linux, what are the plans for AIX going forward? We"ll continue to invest in AIX for as long as we have customers that are using AIX. I still have a team on OS/2 by the way.

I suspect it"s not as big a team as it once was. It"s a much smaller team. You can certainly speculate and say, "Well, gee, if Linux did everything that AIX does and everybody liked Linux, you might over time see the interest in AIX wane, and that interest get picked up by Linux." It"s all speculation. We"ll have to see whether or not that happens. Certainly, if it does, it"s pretty logical then what we would do. We produce AIX to drive our pSeries hardware business. That"s why AIX is important. It adds value to that hardware platform. It makes the hardware platform go. If something else makes it go, we can switch to something else.

But you"re happy to leave Linux distributions in the hands of outsiders? Yup.

You don"t feel the need to have more control over that? No.