IBM's top Linux expert

Question: Irving, your title has changed since you last talked with the Linux Executive Report. You're now vice president, Technical Strategy and Innovation. What new responsibilities have come with this title?

Answer: First of all, as the title suggests, I look at the overall technical strategy across the company-what we're doing in systems and software. But what I'm spending much of my time on is the major technical elements driving services and solutions.

This is one big area of major change. I'm particularly focusing my time on the marketplace where some profound things are going on. The marketplace is where our constantly advancing technology meets market changes and evokes new business demands for incredibly sophisticated solutions. This is where the greatest innovation is taking place.

Incremental changes are more orderly. People understand what they are-both the people building them in our labs and our customers. The drastic changes are a little harder to understand.

You know, like when the Internet came along-what does it mean? Or when Linux or on demand came along-what do they mean? So, I'm particularly focused on some of the more profound changes that are generating major innovation in the marketplace.

Q: How do you define what IBM's innovation initiative means to customers? How do you see Linux fitting into that picture?

A: In a way, innovation means that we help our customers exploit the best possible technologies to get the maximum business value. We help our customers do what they need to do to be competitive and-hopefully-leaders in their industry by helping them take advantage of IT.

Now, Linux has a major role in that effort because it opens up new opportunities. For example, the world of inexpensive supercomputing based on clusters and blades has totally grown up around Linux.

The opportunity to use large numbers of inexpensive microprocessors as clusters to solve new problems in engineering, healthcare and pharmaceuticals has emerged all because of Linux. A big part of pervasive computing consists of embedding microprocessors complete with operating systems and applications into the physical world.

You find them in everything from automobiles and oil-drilling rigs to medical equipment and consumer entertainment devices. Much of this is enabled by Linux and many infrastructure applications are based on Linux as well.

So, when I think of Linux, I sort of think of it the way I think of TCP/IP; they both enable lots of new products to come to market. People don't have to think about what's the right networking protocol-they use TCP/IP. Likewise, people don't have to think about what's the right operating system-they use Linux.

Q: How much would you say Linux has gained in maturity since you first led the initiative at IBM?

A: Well, we started the major Linux initiative in January 2000; that's when I came to work at the Systems Group for Sam Palmisano to run our Linux initiative. At the time, Linux was like a young rookie with incredible promise and we really liked how good the community was.

The reason IBM embraced Linux was because our top people told us this was really good and it had very big potential in a number of areas. But at the time, Linux had its limits. It ran on 1- or 2-way systems, for example, and had very few applications on it.

The picture today is dramatically different. Linux is a very good, reliable operating system up to 8-way SMPs (symmetric multiprocessors) and is even able to run on 16-ways quite well. And given the power of microprocessors today, an 8-way SMP is a pretty powerful system.

Linux has improved immensely in terms of reliability and security; it's a very good commercial system. I mean, it isn't quite up to AIX, i5/OS and z/OS, especially on large SMPs, but it does incredibly well.

The support for Linux has gotten much better across the world; for example, the number of people who are trained on Linux is much, much higher, the ecosystem of ISVs and channel partners that support Linux has increased by leaps and bounds. So, Linux today is dramatically more mature than when we started in January 2000.

Q: What are you hearing from customers about its value, growth and future?

A: You know it's very interesting. When we first started, we had to convince customers that they should pay attention to Linux because it was going to be very important. By 2005, you have to be somebody in a very obscure portion of a very obscure country not to have heard about Linux.

I think customers accept Linux as a strategic force in the information-technology industry. They may wonder where it best fits, given their unique requirements, but the conversations are very different than they were at the beginning.

We've spent time with our customers, helping them understand the value of, let's say, Linux versus AIX and when one is better than the other. We've spent time with customers contrasting Linux versus Windows and clarifying when one is more appropriate than the other.

Linux is now a conventional element of the normal development of their infrastructure. It's accepted by an increasing number of our customers as a serious building block of the infrastructure.

Q: How do you think Sun moving Solaris to open source will affect the Linux momentum?

A: That's a very interesting question. In my mind, the key to an open-source project is the quality of the community behind it.

When we at IBM choose to join an open-source community, whether it is Linux, grid, Apache or-more recently-the Geronimo community that's building an open Web application server based on Apache, it's the quality of the community that drives the decision. How good are the people? How diverse are they? What kind of resources do they have? The better the community, the more likely we are to join it.

Now that's very different from a vendor saying they're going to let other people look at their source code. Because, by and large, the only people interested in a proprietary product are the existing people who work with it. And even then, the product wasn't designed to be worked on by people outside of the company.

So it's difficult to look at that. That's a long way of saying I don't see the Solaris move as coming anywhere near to the community that Linux has built up. It'll probably be of value to some parts of the Solaris ecosystem, so it's probably a good step for Sun, but I honestly doubt that you'll see the large numbers of smart people in other companies and universities help build this offering like they do with grid, Apache and Linux.

Q: You've referred to the 'commoditization' of the operating system. What implication does that have for Linux, software developers and users?

A: I would say, rather than commoditization of the operating system, that there are a few components that are fairly standard; people don't even think about it-that's sort of the starting point. TCP/IP for networking is one of those standard components.

It's hard to imagine a product that needs to communicate that doesn't start out with TCP/IP. If people are building some kind of Web server, Apache has far and away the most popular following, so people always start with Apache as the Web server of choice.

Now what we are seeing is more and more that-everything else being equal-people will pick Linux as their starting operating system for their new application, especially a brand-new business that may be leery about license fees and royalties.

Plus, Linux is supported by a large ecosystem and you have people learning about it in universities around the world. On top of that, Linux is very inexpensive. You essentially only have to pay for support.

Therefore, Linux has become part of the standard that's taken more or less for granted, especially for many new products, businesses and vendors. You may wind up saying, 'OK, for this particular product maybe I need AIX, maybe I need Windows, maybe I need z/OS or maybe Solaris.'

But the starting point of the conversation is Linux. So, in that sense, Linux really has become a standard operating system. And the fact that it runs on every single microprocessor and runs on systems from small, embedded ones to nodes in very large supercomputers means that it has wide applicability.

I might add that Linux will even run on architectures that haven't even been developed yet. The implications are huge; that's why Linux has continued to develop a bigger and bigger following. Month by month and year by year, more and more applications become available on Linux and more and more businesses use Linux.

Q: How does Linux migration enable customers to focus more on innovation and companies to become an On Demand Business?

A: Well, I think that by embracing Linux in the infrastructure, customers get a certain freedom. For example, often customers pick Linux because they want the freedom to run it in a small distributed system and later move it to a partition in a large mainframe or partition in a big pSeries* or iSeries* [system].

They don't need to think very hard about it with Linux. They know that freedom is there. That freedom then lets customers focus their energies where they are needed: on customers, the right business processes, greater efficiencies and the right business model.

Q: What kind of influence is Linux having on grid computing?

A: Linux has a very big influence on grid computing. First of all, a lot of the same communities that are at the leading edge of grid computing-the world of research, universities, supercomputing-are the same communities that have embraced Linux.

So there is a natural affinity for grid to be built on Linux. Even though one of the powers of grid protocols is that they run on every single operating system, they don't need to run on Linux. Then, like Linux, the grid protocols are developed in an open community.

So again we have very smart people from around the world in a community collaborating on development. It's the same powerful model that's so appealing in the Linux community and it's being applied to the grid community. I think that over time we'll learn what kinds of changes to make in Linux to better support grid computing, especially some of the virtualization requirements of grid computing.

We'll see features like the POWER hypervisor moving into Linux or even related products. So, there's a lot of affinity both for cultural reasons as well as for business reasons.