Excitement was in the air. It was the turn of the century and the dot-com boom was in full effect. The bubble just kept growing and growing, and it was probably invincible! The consulting company that Chris G. worked for had gotten their largest contract yet — the first they'd ever had in the seven figure range. Their goal? Build a sister site for a major investment bank that would serve news and analysis on the latest in the investment world.

Going in, the client made it clear that they wanted to do things right and that money was not an object. This was the height of the bubble, after all. In addition to Chris's company, top consultants from IBM and Sun were hired and involved in virtually every aspect of the project. After the majority of the analysis was done, it was time to discuss the bank's hardware needs.

Sun's analysis, in short, was that they had to go big — no namby pamby rackmounted servers, no weakass four-processor units. No, they needed the behemoth — the E10K.

The E10K was big in every sense of the word; a minimum of four processors (with a cap of 64), housed in a hulking 5.8'x3'x4' case, with a weight of 1,800 pounds. Because of the large user loads they were anticipating, they had to go full throttle — all sixty four processors. This was probably more power than they needed, but at least they wouldn't have to worry about upgrades for a while. Once the site opened, though, it would blow up and the number of users would surge to the levels that would require this kind of power.

But that wasn't all. The system defined three classes of users — registered, unregistered, and staff. Sun recommended one E10K for each user class.

Each E10K with the proposed configuration cost around $2,000,000. So thus far, they were looking at six million dollars. Oh, right, they need a BCP environment, so double it. Six servers, twelve million dollars, over five tons of hardware.

Of course, the environments for dev and QA should perform similarly to production, so toss on another six E10Ks each.

If you've been following along, you'll see that it's up to 18 servers. Thirty six million dollars worth of hardware. When Sun presented the total for the servers, some additional hardware and fees brought it up to an even $40,000,000.

The price tag wasn't that hard for the bank to swallow, however. It had been running an aggressive marketing campaign to drum up interest in the new site and they were sure that buzz was building. Existing customers had the exciting opportunity to pre-register for the site so they could have all of the benefits on day one. After the multi-million dollar campaign, it was time to tally up the pre-registrations.

One-hundred and six people.

It wasn't the end of the world, though - it could still work if they somehow managed to squeeze hundreds of thousands of dollars from each registered user. OK, maybe it was the end of the world. Realizing that there was no way they'd ever be able to recoup the money spent on hardware, software, consulting, and ongoing maintenance of this system, management canceled the entire project, sending nearly a floor of developers to the curb and putting Chris's company in ruin.

But the good news was that the industry learned from this incident, averting similar "money is not an object”-type spending and keeping the dot-com bubble invincible to this day.