Stratus is one of the oldest, and possibly one of the most obscure, computer companies in the industry. Known for continuous computing, it boasts that its servers never go down. We spoke to CEO and President Dave Laurello to find out how the company has managed it and how it plans to survive the challenges of the 21st century. Part 1 of the interview is here, part 2 follows next week.

Q: How has Stratus survived 20 years or more where the competitors have disappeared?
A: We're one of the original Massachusetts IT companies, and we're coming up to our 25th anniversary. How we got there is through an IPO in '82 with two basic divisions: telcos and financial services. Eventually, after Ascend's acquisition of Stratus and Lucent's acquisition of Ascend, we bought the company back. We're privately held though we report publicly. We may appear quiet but we're not going to market our way out of that.

Instead, we're going to focus on certain segments and dominate them, and have no aspirations to be the third or fourth computer company in the world. We have a very strong, healthy legacy computing business that's very loyal and we want to retain that loyalty.

That base provides a financial foundation to invest in new technologies, including the ftServer line. One of the issues we've been able to overcome is that Stratus was known as a company that was very expensive and not competitive from a price/performance point of view. Great availability but the technology was two- to three-years old. Now we're back on the Intel price/performance curve and we're very competitive compared to clusters. We have the same level of availability, if not better, than we had 20 years ago.

Couple that with the market which has undergone a general shift, with people needing availability. In this whole online world, more customers can measure their system downtime as a function of revenue loss. The more that people have that mentality, the easier the sell. Everyone talks about it but we feel we invented it - we're the only provider of a continuous, open platform with a unique service model, which allows us to manage customer systems remotely. It's about keeping focused.

Q: Wasn't it risky abandoning VOS and moving to a mass market model?
A: We're just migrated VOS to our open platform and will continue to sell to our VOS base. But the market has decided that new applications don't get written for proprietary platforms - they have to be Linux or Windows. There's nothing we can do about that. It's actually a lower risk path than staying with existing OS but we're not abandoning that, we're providing a migration path.

It's not a risk, just a natural place to go as we ride the coat-tails of the growing economy.

Q: How have customers reacted?
A: Since we started shipping the ftServer servers in 2000, we have about 1,000 customers in that 3.5 year period of time - more than we've ever had. However, the cult following that we've had with our older VOS operating system isn't the same. At the shows we have, they keep coming back and they want to talk to Paul [Green, senior consultant] and will go into the depths of the OS.

Not many systems companies have the abilities to go into the depths of the product like that. Other companies pull pieces together from different areas of the globe and all they know is speeds and feeds.

Q: Given you've been used to a small group of loyal customers, how are you managing customer growth?
A: Customers see our system as differentiated and we don't get compared with Dell, HP and IBM on the commodity level. We're a very services-led culture. If a customer calls our 24 by 7 customer assistance line, a human being answers the phone, not a call centre someplace. If they have a problem, we don’t say 'prove the problem's ours then call us back'. We just start debugging because we're experts in all the layers of hardware and software in the box. We're told we're a step above that of the others. They don't get a person but an automated switchboard message, and they see us as a specialised, expert company and the value that. So far it hasn't been a real challenge.

However, we now do more business through the channel so you lose some of that contact.

Q: How's that growth going to affect product development?
A: Enough of our business has direct touch so that that hasn't been a problem and our partners provide us with good feedback.

Q: You don't own Windows, so how can you get the level of understanding of the system that you claim?
A: We've worked very closely with Microsoft from the beginning, so they've put hooks in their code to work with our platform. It's in the standard Windows system. If we need help from Microsoft, we have a close relationship and they help us.

Q: When you say hooks, are we talking about unpublished APIs?
A: No, changes in the kernel to take advantage of our platform and our intellectual capital. It's on the standard CD that everyone else has.

If you think about where we're going, we took continuous availability and migrated it to an open platform. Microsoft and Intel saw the value to them of taking their products into areas they hadn't been before, such as Unix clusters in the public safety market - we dominate that now in the US. So they were happy to help us provide that solution.

Q: Can you give an example of the kinds of changes to Windows that we're talking about?
A: No.

Q: Even at a higher level?
A: Our system's differentiator is that if the application fails, the system continues to run, unlike a cluster where there's failover. Then when you reinsert the component - hot-swap it - the new component's got to come up to speed to where the old one was. That re-synchronisation happens in microseconds, and the reason that happens is because of work that Microsoft has done in the kernel to work with our IP.

Q: What's your relationship with Intel?
A: Intel is an investor in Stratus. It invested before we'd shipped anything because they saw the value in the deterministic systems that we were developing. It's like their chip test line: they need it to give the same answer every time. One of the things they always worry about is 1+1=3. And we test the fringe, the corners, some of things that they may not normally test or that customers don't care about.