IBM is one of the strongest advocates of the Linux, and has invested billions of dollars to ensure its systems support the open-source operating system.

In the mainframe world, IBM has shown the potential of Linux as a system fit enough for the toughest corporate task, while Linux has provided 'big iron' with some much-needed street-cred.

As worldwide vice president of Linux sales, Carol Stafford is charged with the task of spreading the word of Linux to IBM's biggest customers.

Stafford talked exclusively to Computing about the importance of the operating system for the world's largest IT supplier.

How did you first get involved with Linux?

I have been working in Linux and especially Linux on the mainframe since 1997. It was very exciting. I was working with the IBM engineers in Germany and they were running Linux on the mainframe without any operating system underneath it and developing it from there.

You came from the mainframe world - did it take time to understand what Linux is about?

It seems obvious that you can have the mainframe working in the background and you take all the strengths of Linux at the front-end and you tie the two together. But when I started it was black and white - Linux or z/OS [IBM's mainframe operating system], one or the other.

I can remember when I started, one of the executives coming up to me and saying, 'You know Carol, I am not sure if you are part of the answer or part of the problem.'

Now what you are seeing with the iSeries range is smaller mainframe boxes that have the same value proposition as very high-speed, Linux virtual servers with a very sophisticated database back-end. So our customers are seeing huge improvements in performance in big applications like SAP. We have customers seeing 60 per cent throughput improvements in big batch jobs.

How successful is IBM's Linux business?

Linux is growing hugely - in 2003, for IBM, almost 80 per cent growth year on year. In the first quarter 2004 it was 92 per cent. I haven't see the second quarter figures yet but we have revenue projections that we are far over-achieving.

Something like 80 per cent of our customers have Linux projects of some sort going and we ourselves have 3,500 production projects going - everything from web portals, to big transaction SAP stuff.

Our Fishkill facility in up-state New York is a new fabrication plant for our Power chip technology - it uses Linux from top to bottom.

All through research, development, fabrication, robotics, purchasing - all run with Linux.

We are seeing significant mainstream pick-up in the enterprise market, particularly in the financial services sector and retail.

We are working at making our technology through Linux and the Power architecture all-pervasive. So you see it in the games arena in Tivo set-top boxes and Sony Playstations and you see it in embedded devices. It's being used in the Mars Lander for example. Linux is integral to our grid technology.

How is grid computing doing as a business?

It's more understandable to people now and we are seeing it particularly in scientific and research applications. Around the world, research labs can grid their machines together. When I was in Finland recently I talked to the Computer Research Institute. They have just bid for and won a research grant from the EU on a project with no compute power at all because they are able to grid into other research departments around the world and use their power. It gives a level playing field.

Grid is helping Linux in new markets. The emerging market story is huge and that is pushing Linux into the desktop arena. You can see that in places like China where they have this huge software market - $35bn in the next couple of years, according to IDC. They are endorsing Linux and paying stipends to the corporations there to use Linux, in an effort to try and keep the software business inside China. They believe that Linux is an equalising forc e with the Western world. They can get people highly educated in Linux - that it is something they can easily mobilise their universities to do - and it will bring work so they can compete with India and the like.

Do you see Linux as being the operating system - that there is really no need for anything else?

We don't project that at all. We project continued growth. It's the fastest growing operating system and it is growing shipment share, where Microsoft is losing shipment share of servers. So we think it will probably get equal to Microsoft in 2006.

We don't believe in the near future that it will take over things such as z/OS in the database, large transaction area. We think it will add superior capability to the front-end of those. It will allow them to do more things.

So you disagree with Microsoft that Linux is only a server product?

Our strategy had always been the server. But what we are seeing this year is that our customers are forcing us into the desktop. They are trying it, they're doing it. IDC thinks the Linux desktop will grow by 30 per cent per year. We have 30,000 Linux desktops in IBM. What we are telling customers that want to do is to segment their users. Some people can easily use Linux, particularly single-function users - people such as bank tellers and weather forecasters.

We have people who want to isolate their users and only have them do email and one or two functions. Those are easy to change. If you have major users of Microsoft Office, what's the point? In IBM we think that by the end of 2006 we might have 50,000 or 60,000 Linux desktops.

There are a lot of options now for desktops and we will have our own Lotus Workplace, which is a server managed client.

Cost used to be the major factor in moving to Linux - is security now more important?

It used to be that people looked to save money and some saved a little and some saved massive amounts. There are two things that are important to the enterprise customer: one is security and the other is performance.

The argument here is that once it is pervasive you will get more hackers. Yes, there will be more hackers, but there are two things that are different. First, it was built in a way that security was already thought about. It is smaller, so it is easier to fix. Secondly, in the open source world, most of the problems are diagnosed in about 30 minutes. So something might happen here and then someone in the US or someone in India will see it and fix it. It doesn't have the same problems as Microsoft.

You think about the cost of having to keep the latest updates on for a security fix. Particularly in Europe, people are very aware of this -a lot more people talking about it over here.

Performance is the next key. It's much better but then again you think about the amount of code running in the background and it is pretty small.