I was a corporate pilot for an accounting firm in Reno, Nevada. On Monday mornings I'd fly a group of CPAs down to Las Vegas, and then I'd sit in a house all week long and wait for them to get done. They gave me an Osborne portable computer to tinker with. I taught myself Basic and wrote a program that tracked the operating expenses of the airplane: who flew where, how much fuel and oil we consumed, and what the various maintenance costs were.

As a result, the CPAs realized that they couldn't afford the airplane, and so they let me go. "But you're pretty good with computers," they said. "You might want to think about changing careers."

I went from selling computers at a store in Reno to developing applications for my customers. I started a consulting firm in the Reno area, and shortly afterward Reno Air asked me to build its IT department. Then a company out on the East Coast called Independence Air recruited me to do the same thing. That airline went out of business, and JetBlue recruited me to bring some of the automation that we'd done at Independence into the JetBlue IT environment.

We built Independence Air with a shoestring budget and just a handful of people. It was some of the best airline technology that was out there at the time. I had a good group of people working with me, and I brought some of them with me to JetBlue, where there's even more opportunity for innovation. We've got a lot of new technology in the works and a lot of it is Microsoft-based.

We're mostly a Microsoft® shop, though we do have some implementations of Oracle, SAP, Cisco, and Juniper for various purposes. We do a lot of .NET-based development with our own service-oriented architecture that we've dubbed the JetBlue Framework. That approach allows us to integrate all of our disparate systems and enforce business logic so we have a defined set of metrics that we use to run the business. It improves our programming efficiency and the accuracy of our data.

What's the total cost of ownership, and what benefit does it have for my customer, whether internal or external? One time when I was working for a previous employer, a vendor told me that his product was 12,000 input/output transactions per second faster than the competition. I said, "That's great. Will my end users see the difference?" He said, "Well...no." You can go out and buy the biggest, fastest, greatest thing, but in five years that thing is going to be a boat anchor.

If we're using Linux at all, it's very specialized. In the networking world there are better tools available on Linux to accomplish certain things at a given price point. But we don't have a large implementation of Linux by any stretch of the imagination.

We were starting down that road, but it quickly became apparent that our talents are in the .NET world. I know from previous experience that you can't just take a Linux class and become a Linux expert. So I kind of pulled back on that so we could focus on what our core skills were.

Some people bash Microsoft because of all the security updates that come out. I can tell you that we were getting as many, if not more, security patches for Linux. It became apparent that Linux really isn't any less expensive to operate than anything else, and if you don't have the skills in-house to support it, it can actually be much more costly.

It's the same thing: if you have the in-house resources to support it, Linux can be as reliable as you want. But I think that Linux is more prone to tweaking and tailoring than typical Windows environments, and that has a direct impact on reliability. Linux guys tend to have their own tool sets and scripts that they use, so it's easy to get inconsistency from server to server. If you have a mature, disciplined IT shop with very strong change control and it's really enforced, you've gone a long way toward preventing problems in that area.

You need to understand the business around you. IT is of no value if you can't apply it to the business. The most successful IT people that I know understand that you don't just throw technology at a problem; you've got to have a process to solve the problem and apply technology to the process.