Skip to main content

Temperature Aware Storage Modeling

We worked to reduce storage energy consumption by employing temperature aware data placement and migration. Since heat removal (cooling) from storage devices is an energetically dominant activity in data centers, and since disk operational temperatures (working temperature band) is limited, we realized that controlling data placement within the storage may eliminate automatic triggering of some cooling fans. Hence, by properly migrating data (per storage workload patterns), one can reduce certain temperature peaks and thus avoid turning on some of the cooling mechanisms. Consequently, storage system energy consumption may be reduced.