Looking Forward: Power, Cooling, and the Data Centre
Where energy efficiency meets cost efficiency
In the future, we may look back on 2006 as the year that power consumption, cooling, and energy efficiency in the data centre ceased being a back-burner issue for IT and facilities managers and elevated itself to become one of the forefront, if not leading, issues for many.
While those “in the know” have always been aware of HVAC and power distribution limitations, until recently it has not been a noticeable issue. During the past several months, we have seen vendors focus on the issue of energy efficiency through various initiatives including HP’s Smart Cooling , EMC’s Energy Efficiency Tool, Sun’s Cool Threads, the latest Energy Star Specification, and VMware’s energy utility rebates. With the competitive attention now being brought to bear, we expect to see this topic remain at an elevated level during 2007 as vendors line up their competitive differentiation and definitions of what exactly energy efficiency is all about.
Although much of the cost cutting and resource gutting by CIOs and CFOs during the first part of the twenty-first century focused on infrastructure consolidation and headcount reductions, it didn’t take too long for the impact of $75/bbl oil and 22¢/kwh electricity to reach into the data centre. At the same time, rather ironically, all the focus on server and storage consolidation combined with ever denser form factors such as blades has changed the heat generation and dissipation characteristics of the data centre. Thus, the limitation of physical reality has once again impeded progress in our collective journey to a virtual IT existence. Yet there are many similarities and lessons to be drawn from the “consolidate, simplify, and virtualize” mantra of the past few years. Just as inefficiencies in server and storage utilization have led to consolidations featuring closely monitored virtualization schemes, we are now witnessing the same opportunity with cooling and power consumption.
Over-provisioning of cooling and power is inherently just as inefficient as over-provisioning anything else. If machine rooms are continuously cooled to meet peak loads, then a lot of kilowatt-hours are going to waste. Likewise, if the actual power being drawn by equipment is less than the wiring supports and designed to a worst-case scenario that is unlikely to be achieved, there will be unnecessary breaker panels and conduit being installed. From a financial and operational perspective, targeting the cooling where it is needed, only when it is needed, just makes sense; anything emitted beyond this is simply waste that impacts the bottom line of the business. Similarly, electrical circuits that are underutilized represent an underperforming investment.
Sponsored: Benefits from the lessons learned in HPC