>"There will be more to come, but the art of data center design has finally gotten around to addressing its major sources of wasted power. "We may move one day

>"There will be more to come, but the art of data center design has finally gotten around to addressing its major sources of wasted power. "We may move one day from 1.07 to 1.06 or even 1.05," he said. But there will be no more announcements of 38% power savings in a new data center versus the one that came before it, he said. Today is the time for dramatic gains. Tomorrow's gains will come much harder." --- The future measurement will no longer be PUE, but compute density in a given square footage. Especially when you have single service facilities like Facebook, and standardized architecture offerings like Amazon. However - what we hopefully will see is "Designed for the compute cluster" chassis offerings from the major vendors, based on the work of Google, and now Facebook's open compute project. The primary issues we have with DC design, and that are addressed by the open compute model, is that you have all these various vendors; PDU, UPS, Power Bus, Rack, Server, Cable Plant Components --- All these things "fit" together, but they are not DESIGNED together. Meaning that they all get designed in a relative silo, though around standards that make them connectable, but not to become something greater than the sum of their parts. Systems like hot aisle containment are still built effectively ad-hoc and custom for each datacenter.