Data centers are the backend for a large number of services that we take for granted today. A significant fraction of the total cost of ownership of these large-scale storage systems is the cost of keeping hundreds of thousands of disks spinning. We present a simple idea that allows the storage system to turn off a large fraction of its disks, without incurring unacceptable performance penalties. Of particular appeal is the fact that our solution is not application-specific, and offers power-savings for a very generic data center model. In this paper, we describe our solution, identify the parameters that determine its cost-benefit tradeoffs, and present a simulator that allows us to explore this parameter space. We also present some initial simulation results that add weight to our claim that our solution represents a new power-saving opportunity for large-scale storage systems.