Wireless data centers could be faster, cheaper, greener

Cornell computer scientists have proposed an innovative wireless design that could greatly reduce the cost and power consumption of massive cloud computing data centers, while improving performance. In the "cloud," data is stored and processed in remote data centers. Economies of scale let cloud providers offer these services at far lower cost than buying and maintaining one's own equipment. But data centers with tens of thousands of computers draw tens of thousands of kilowatts. "Reducing power consumption would not only cut costs but would be a benefit to the environment," said Hakim Weatherspoon, assistant professor of computer science. Weatherspoon; Emin Gün Sirer, associate professor of computer science; graduate student Ji-Yong Shin; and Darko Kirovski of Microsoft Research have prepared a feasibility study for what they call a "Cayley Data Center," based on wireless networking and named for mathematician Arthur Cayley, who laid out in 1854 the mathematics they used in their design. Their proposal is available in the Cornell eCommons information repository and will be presented at the Eighth ACM/IEEE Symposium on Architectures for Networking and Systems Oct.