Charlie Bess, an EDS fellow, recently wrote in his blog about a new option for pricing utility computing and it got me thinking about 3tera’s pricing model. When we were deciding how to charge for our service we specifically tried to stay with a model people understood, servers, instead of introducing a new metric. As we prespare to go to market I still think that was the right decision and I’ll share a bit about why.

Although people have been talking about utility computing for years, what’s really been sold is a bundling of services; hosting, contract engineering, business consulting, and support. That is, until recently. Sun and a number of startups, including 3tera, have begun to roll out services that allow you to actually purchase computing power without the labor component. Further, several startups are building hardware and software systems that, for the first time, are designed purposefully to enable utility computing. No two of these systems look even remotely alike. Given that variety, I feel safe in saying that utility computing as a catagory is about to undergo a transformation. Not just the billing metrics, but the very nature of what’s sold and how it’s used will change. Therefore, I believe trying to predict what the ultimate billing metric will be before seeing what’s being cooked up in the labs and how customers use it is pointless. Or worse, it could be very detrimental.

How? I’ll explain with an example. About a decade ago I was with a networking company acquired by a phone switch manufacturer. My company had been building data communications gear that was used in the enterprise, so billing for traffic, of course, was not an issue. I’d never even seen a phone switch at the time and after the merger I befriended the “voice” side of the business and was astonished at what I found. The systems were massive, but had little relative throughput. Worse, their cost per bit of throughput was orders of magnitude higher than ours. The dirty little secret, as I learned, was that the built-in billing system costs more than the switching technology that made the call possible. That’s right, it cost more to bill for the call than to provide it. A decade later, AT&T still sends me a five page itemized statement to collect their $7 from me each month. What a waste!

So, while I’ll keep revisiting the billing decision periodically, my thoughts are focused on how best to provide utility computing and how to drive the cost of the service down.