As part of ongoing research into the energy efficiency of data centers, Microsoft Research (MSR) have released a paper discussing the results of testing disk subsystems usage and potential to 'spin down' disks when they are not being used. In the past there was an assumption disks couldn't be spun down due to continuous use, however MSR found during their week long trial on 36 volumes in their data center there were "significant" periods of idle time.
Their solution to helping reduce energy used by disks, up to 45-60%, was to enable functionality called "write off-loading" - this allows write requests on spun-down disks to be temporarily redirected to alternative persistent storage (such as NVRAM or flash).
A few interesting facts from their paper:
enterprise storage sub-systems don't have advanced power management capabilities (because it is inherently difficult) and they consume approximately 10w when idle;
write off-loading is a block level, transparent process;
the key challenge was consistency - "Each write request to any volume can be off-loaded to one of several other locations depending on a number of criteria, including the power state and the current load on the destination. This per-operation load balancing improves performance, but it means that successive writes of the same logical block could be offloaded to different destinations. It is imperative that the consistency of the original volume is maintained even in the presence of failures. We achieve this by persisting sufficient metadata with each off-loaded write to reconstruct the latest version of each block after a failure."
They tested two types of off-loading: machine level off-load (writes are offloaded to a "logger" on the same server) and rack level off-load (writes are offloaded to any "logger" in the rack) and found the latter to utilise the least energy (55% of the baseline).
Check out the complete paper outlining the details of the tests and associated research here: http://research.microsoft.com/camsys/paper-final.pdf