The Uptime Institute's Data Center Design Charrette identified many incremental improvements in data center cooling and energy efficiency, but couldn't reach a consensus on big-ticket gains. In a summary of the charrette's findings, Uptime executive director Ken Brill said the institute had hoped that a "silver bullet" might emerge from the event, with DC power distribution and free cooling presenting the best chances for major progress on the power and cooling challenges created by high-density servers.
Instead, Brill says, the discussions among leading specialists and vendors identified factors that prevent either DC power or free cooling from working as a broad-based solution. Brill writes:
"Alas, the primary conclusion we seemed to come away with was that, based on what we know and on the functional silos within user organizations and between users and vendors, there are no silver-bullet solutions; at least not yet. Nothing is so obvious as to demand immediate inclusion in the design/planning repertoire. We did not find significant site infrastructure CapEx reductions (i.e., 75 to 90 percent cuts) and we did find that some of the alternatives considered involved some major increases in IT reliability risk."
In many organizations, even the incremental improvements identified by charrette participants will be difficult to achieve without an "energy czar" to overcome bureaucratic resistance to new approaches. Lacking that, "little improvement can be expected," Brill writes.
The charrette's outcome underscores the difficulty of the power and cooling challenges facing data center operators. The charrette, held in Santa Fe on Oct. 28-30, brought together 300 of the smartest technical experts in the data center industry.
The event's inability to produce a consensus also leaves data center watchers to wonder how much progress can be expected from industry-led efforts. The Green Grid has focused its efforts on developing industry metrics that can help data center operators get a better handle on efficiency issues. Common metrics would make it easier for data center operators to measure internal progress, and enable industry-wide comparisons that could make it easier to understand which techniques are working well in similar data centers.
But expectations of a near-term technological breakthrough on power and cooling issues may be overoptimistic. "Perhaps our collective hopes for Charrette were unrealistically high," Brill writes. "If things were all that obvious, they would already be happening."
Some new techniques and technologies were identified that can help data center managers improve their efficiency (we'll be writing more about some of these), but some key areas of potential progress have proven too complex for a unified approach.
One of these is DC power, an area where the charrette's organizers had hoped for progress. But the discussion identified many "critical-path barriers" to the use of DC in large-scale deployments, including lack of standards and DC-ready products, and lack of proven ROI calculations to win executive support.
"These are very serious obstacles which may take many years to overcome, and there is no natural constituency with a vested interest to work on resolving them," Brill writes. "In the absence of a much more compelling business case and of broad availability of all types of IT equipment with DC voltage inputs at comparable or lower prices to hardware with AC inputs, UI predict that few owners will be willing to invest the time and resources to go with an all DC system."