* The complexity of data center placement
Land went fast in parts of the Pacific Northwest a short time back. Big companies like Google, Yahoo, and Microsoft bought territory near abundant sources of hydroelectric power in the hope of both guaranteeing, and controlling the cost of, electrical power for their data centers. The fact that this is a region not prone to large-scale natural disasters like hurricanes is a bonus. Presumably, on other continents and in other parts of North America, other companies are making similar acquisitions for similar reasons.
Data center placement has become a much more complex endeavor in the last few years, thanks to the convergence of two factors: the growth in demand from super-dense, super-hot data centers, which we have explored extensively here, and the loss of trust in the power grid.
Data center architects don’t and shouldn’t trust the U.S. power grid in the same way previous generations of designers could and did. There are many reasons not to. Supply is having trouble keeping up with demand in several important regions – in New York and California, for example – leading to rolling brownouts and even blackouts in peak load periods. The power infrastructure is aging and so failing to become significantly more efficient. Key components in the power generation and distribution grid are attractive targets for disgruntled fanatics of many stripes. Natural disasters and climatic change are putting strains on it, whether in the form of outright destruction as in Hurricane Katrina, or in the slow push-over–the-edge of longer than usual heat waves causing not spikes but unsustainable plateaus of energy consumption for cooling.
The banks of the Columbia River beckoned as one potential solution to the problem for Google et al, and other hydro sites will too, but large enterprises, governments, and data center companies should not limit themselves to looking for help from flowing water only. Especially if oil prices stay high, other options will become steadily more attractive too.
Consider, for example, the possibilities of wind turbines in north central states like Minnesota, North Dakota, and South Dakota, where a stable mixture of coal, oil, gas, and nuclear plants is being supplemented gradually with wind power. The local flavors of disaster are floods – easily avoided – and tornadoes, the dangers of which can be minimized by underground construction. As an added bonus, for several months in the year, waste heat from a data center can be used to heat offices or nearby buildings!
Or consider biomass conversion. With new techniques like thermal depolymerization, the potential exists to build plants that will convert organic (in the sense of carbon-based) matter, anything from food waste to cow manure to plastics, into petroleum and natural gas. A far-sighted company might strike a deal with a waste management firm, or a major feedlot or food processing plant, and the local utility, to get a local loop going – garbage to oil to power to FLOPS.
Whatever the solution – whether the flows tapped are of water, air, or garbage – the key is to explore the possibilities and look farther ahead and farther afield. We may run out of dammed spots, but the wind will always blow, and the chance to turn cow flops into teraflops is worth looking into.
