A few years ago when I met B V Jagadish, the man who pioneered the idea of server farms, he had told me about how Exodus, his web hosting company, had contributed to serious power shortages in California in those early days of the dotcom boom. With the proliferation of data centers since then, asset managers and vendors alike have been constantly trying to overcome two related problems—the physical area occupied by servers and subsequently the quantum of power required to keep data centers functioning 24/7.

The arrival of blade servers solved the real estate problem to some extent, though blades took a while to catch on, particularly with Indian users, due to their high costs. However, power consumption continues to be a pressing concern. Today, the cost of hardware by itself is not so much of a headache as is the power consumed by data centers. In a few instances, power utility bills even exceed the investment made on hardware. Increasingly, expenditure on power accounts for a substantial portion of the IT budget.
With the increase in processing power, data center temperatures are hitting the roof. Cooling these data centers means further increase in power consumption. The cost of a unit of power has also gone up. Given the existing demand-supply scenario, it can only go up further.

The good news is that blade servers have continued to improve in features. The new generation of blades is more sophisticated and powerful than its predecessors. At the same time, vendors are also working hard to bring down the power consumption of data centers.

In today’s context, both blades and virtualization should greatly contribute to resolving the issues of power and cooling. Before you decide on a vendor or a systems integrator, gather up-to-date information and do a thorough RoI analysis. Or else, those savings on power and cooling could remain elusive.