Dear Nathan,
You're insane. There's no way the internet bubble is going to burst. To (mis-)quote Bill Murray in Kingpin, "We're on a gravy train with biscuit wheels!"
Yours truly,
Joe Blow, MCSE

I hate to be the harbinger of bad news to the multitudes, but I think we might have to face the fact that (don't quote me on this) some Internet start-ups might go under. Shocking news, and you heard it here first.

Over the past several months, it's become obvious that investors have lost faith in many technology companies, which has lead to a kind of down-the-food-chain string of lowered earnings, lowered stock prices, and even layoffs. I don't pretend to be a financial expert. In fact, my idea of sound financial planning is to look at my checking account balance once a week or so via my bank's website to make sure that I haven't overdrawn yet. One thing I do know however, is that free stuff is cool. This perhaps is the cornerstone of American culture. Free refills of drinks at restaurants, free samples of shampoo in the mail, the long-since lost free disks that I used to get thanks to the generous folks at AOL. Everyone likes free stuff, it's in our genes or something. Why then is it so hard to convince management that free, open source software is a good thing? I'll come right out and say why- they were spoiled.

That's right, I said it. Management got spoiled by the internet boom. The money was flowing in, why waste any time considering those silly "free software" solutions when you've got perfectly good software that you can buy? Now, the boom is over, the economy is finding it's equilibrium, and management is suddenly realizing that they don't have an endless supply of money for software licensing. Many companies are trying to dramatically cut back on spending, so that their profit margins are better. It seems, however, that far too many companies don't realize how much money they could save by using open source software where applicable.

Now, I'll probably get flamed for saying this, but believe it or not, my biggest beef isn't with Microsoft in this case. Although licensing fees to Microsoft amount to huge sums for many companies, I'm not going to touch the "make everyone in the company use Linux on their desktop" argument with a ten-foot pole. I'm not even going to make broad generalities about how "Linux is ready for the enterprise." What I want to do is explain how myself and a few other open source geeks at my company are managing to make the best of the tech slump by taking this opportunity to push the "free as in beer" aspect of free software.

At the company where I work, we put some serious loads on some pretty serious big iron. This big iron is not cheap, in and of itself, and added to the hardware cost is the enormous cost of the operating system and other system-level software packages. Most of the applications that we run were written in-house, except for some middleware pieces, such as the CORBA implementation that is used. About three years ago, a team leader theorized that we could get significantly better performance from many of our applications if we could distribute the workload across many commodity machines, rather than running on a single monster SMP machine. It was a good idea, it was a perfectly feasable idea, it was a cost-effective idea. Unfortunately, it took two years for him to convince the powers that be to allow him to build a prototype system.

Originally, we had planned on using many smaller Alpha processor-based machines. Unfortunately, after software had been added into the price of the hardware, the cost was the same as one of the big iron machines. At this point, we decided to investigate the use of Linux on Intel processor-based machines. This not only significantly reduced the hardware cost, but also removed the cost associated with operating system licenses. The next obstacle was the CORBA implementation that we were using- not only was it horrendously expensive, the version we were using at the time did not run on Linux.

Enter TAO (The ACE ORB), an open source real-time CORBA implementation built upon the ACE (Adaptive Communication Environment) networking libraries. By getting rid of the closed-source proprietary CORBA implementation that we had been using before, we were able to make modifications such that our software would run on darn near whatever platform we wanted to try out. At this point, we were ready to roll- all we needed were enough machines to put some hard numbers to our theories.

Unfortunately, this was also the point at which the economy began slowing. We sat around on our laurels, telling anyone and everyone who would listen of the amazing things we would do, if we could only get the machines to prove it. Spending was cut back, slowly at first, then more drastically as the economic slowdown progressed. We tried to tell people of the millions of dollars we could save by completing this project, but it all seemed to fall on deaf ears. Then, a kindly hardware vendor allowed us to evaluate several machines for a month. We had the ammo we needed. With hard performance numbers to go along with the hard cost-savings numbers, we couldn't go wrong!

Within a week of getting the machines, we were up and running, and producing results that made even us stare in wonder. We used the same test cases that they used on the big iron systems, and the results were simply astounding. For roughly one-tenth of the cost, we were achieving three times the performance in the big iron's best-case situations, and up to fifty times the performance in the big iron's worst-case situations! In fact, we had almost eliminated the gulf between best and worst-case performance. This was ammo that we could use.

We did a demo for several members of management, and with the numbers we had collected sitting in their face, we were finally able to make them realize just how good for the company it is, in many situations, to not lock us in to old paradigms, just because "that's the way we've always done things." The people present for the demo were so impressed, in fact, that one of the managers had to place a gag order on everyone who had seen the numbers, so that we wouldn't have sales people get wind of it, and start using the numbers in their sales pitches before we were ready to deploy. We're doing a demo for the CEO next week, and we're fairly sure that he's going to be just as impressed as everyone else has been.

At this point, many people are probably thinking, "Nathan, what's your point? When are you going to tell us how to catch the gravy train?" Well, my point is that even in a serious economic slowdown, there are opportunities for inserting open source software into companies. I say "inserting", because that's what it really is. I look upon it as getting the GPL's foot in the door. Once the foot is in the door, it's a lot easier to get inside fully. I have a dream. I have a dream that one day, companies will freely choose software based upon its features, merits, and drawbacks, and not upon the manner of its licensing.

As for the gravy train, I think it left the station while I was typing this. Hmm, maybe I can catch the caffeine train... I think I see it coming down the track now...