I say the best way to know what to expect out of the world's biggest tech companies in terms of what YOU can use in your own operation is to look at the approaches they are taking to their own IT operations. I'm in the 'do-as-I-do' school on green IT policies.
So, it was with interest that I skimmed a new white paper published by Microsoft Global Foundation Services (wow, what does THAT mean?). The article carries the rather promising name: "A Holistic Approach to Energy Efficiency in Datacenters," although it's not a long piece (maybe 6 pages). In it, the engineer warns that looking too closely only at metrics like the Green Grid's Power Usage Effectiveness (PUE) can skew true performance. He writes,
"Just improving PUE should not be an organization's goal. The real goal is to eliminate waste and pack as much compute capability in the available power budget. PUE can be a useful indicator of energy efficiency, but it can also mislead you if used blindly. Take, for example, a scenario in which the fans in a server can be removed without impacting its performance. The elimination of fan power reduces the IT power (fan power is part of IT power), improves energy efficiency, but it also increases the PUE."
Actually, this is another example of why you cannot just look at cutting IT energy consumption without considering the facilities side of the equation for cooling technology. Have you talked to your facilities counterparts lately?
The paper also discusses why data center container and modular designs have been useful for Microsoft (they are used in the company's new Chicago facility) although not necessarily ISO standard containers.
Frankly, I'm not a data center architect, but I can see why Microsoft's experiences, as well as the real-world experiences of companies like eBay or Google that have built their entire business around their data center capabilities are worth emulating.
Some reference materials: