In consolidation you put multiple applications into a single server. In the traditional unix world, that was easily achieved (and still is) because the system is securely segmented.

In virtualisation, you create virtual environments which imitate what an application would get if it were on its own.

You can consolidate into a server which is a single operating environment or multiple virtual environments.

It is all now soooo confusing because of the numerous degrees of separation available: whole server, hard partition, soft/logical partition, hypervisor layer, containerisation and plain old apps sitting next to each other.

Don't mock the benefits of server consolidation using, for example, VMWare.

We are saving tens of MW a year on our data centre power bill by consolidating up to 30 prod servers or up to 60 dev/QA servers at a time onto a single blade server running ESX.

Migrating over 300 servers a year, many that are space and power inefficient old dungers, is providing real bottom-line benefits to us. We are seeing a reduction in the power required both to run physical machines and to cool them, and in physical cabinet space.

Our next objective is to remove some of the fixed air processing equipment for further capital and OpEx savings, leading to a lowered data centre infrastructure cost accruing over the next four-to-five years.

Oh, and not purchase any more servers that run at 3-8% average utilisation.

I'm surprised that no one mentioned Building Automation or Facility Management Systems as a way to cut energy costs, and increase security and system reliability. Forget timers! Lighting control is the EASIEST thing that can be done to save energy but there is WAY more that is possible.

For example the RFID swipe card you use to access the data center tells the FMS to turn on the lights and open the doors all the way to your co-location space but no where else. It has also told the security system to take your picture upon access and record you as you enter and reach your co-lo space. The monitoring and control of the data center temperatures, emergency generator, computer room AC units, and UPS systems by the FMS have prevented data loss, logged voltage and phase issues, triggered preventative maintenance calls all on top of running the data center AC systems at the least possible energy cost. How about Energy optimization strategies like Ice Storage, using least costly power at night to make ice that will provide cooling during the day, or using outside air to provide "free cooling" when OA temps are low enough. Perhaps electrical load shedding could be combined with feedback from processor load info to reduce cooling when server/processor loads are low? Even something as mundane as variable speed drives on chilled water pumps will save energy.

There are lots of things that can be done to conserve energy but you have to have some idea of what is available. It's kind of funny how computer/IT people go all Luddite when you tell them about building automation. "Oh we could never adjust the Liebert discharge temperatures 10 degrees warmer during low load conditions. The servers could crash." (Not bloody likely between 12:30 am and 6:00 am) Or this is rich... "Oh that system sounds too complicated" (reads as I didn't think of it and I don't get to have absolute control of it.) Or this one, "What do you mean you want to have your automation devices running on MY network? I won't stand for it!"