Central electricity management in IT
by
Charlie Bess
In one of the podcasts I listen to they were talking about how folks in the northeastern US and the UK are starting to put natural gas power generators in their homes to generate electricity as well as heat their houses. Any excess power generated can then be placed back on the power grid. These devices can be controlled (to some degree) by the power company. This would allow the power company to fire up these generators on those peak electricity demand days so they’ll not have to build another power generator. The reason why it is limited to the northeastern US is because of high energy costs making it cost effective. The UK is providing tax benefits in addition to their high energy costs.
The IT version (I’d heard proposed by a hardware company) is to put CPU clock regulation under central control and allow the IT organization to turn down the power consumption of the CPUs on days when the organizations power usage is getting beyond a limit. The problem with this perspective is that it implies that the value generated by computing is independent of instructions executed. Those instructions should be generating value to the business. I’m sure there are cases where this could save the organization money, but it could easily be a frustrating experience. It might be simpler just to force all the machines into a sleep state unless user interaction takes place, but that capability already exists. The other interesting issue with this proposal is that the organizations who are the most cash strapped and want to do something like this would need to replace all their hardware, so I doubt that it would ever happen.
There are better techniques to enforce system configuration settings now, so you can strongly encourage people to use the power settings on their existing devices constantly.
