Response from: Steve

This is an interesting option, but it is just the client side of the equation. I would be interested in knowing how much the server side consumes. I assume it scales better per user because there isn't as much overhead, but this post somewhat misrepresents the efficiencies. It's less like getting on a motorcycle, and more like taking public transportation.

Response from: Scott Wilson

Oddly enough, I considered the bus analogy first, but I didn't think it was any better a fit after I thought about it a bit; in most organizations these days, the server capacity is already there, whether you are accessing it with an energy efficient thin-client or a full-on PC. The whole thing is really a crappy analogical framework in the first place; my bad. The point was it's a way of approaching the problem which requires some conceptual retooling to grasp. It's not simply that the server is used more efficiently (although it is) but getting used to the idea that getting things done doesn't require quite the horsepower that most people assume. Getting back to the terrible analogy, the motorcycle doesn't have the horsepower to haul a trailer, but you don't need that just to get from point A to point B. You're assuming the efficiency comes just from a reduction in overhead, but in fact there can be a reduction in overall processing consumption if the system is designed properly. Studies I've seen cited have claimed a 50% reduction, going from 85W for a PC down to 40W with a combined thin client/server. Note that the 85W for the PC doesn't include the server that the PC was probably also using already.