MEMS-based storage devices (MEMStores) are significantly different from both disk drives and semiconductor memories. The differences motivate the queswhether they need new abstractions to be utilized ms, or if existing abstractions will be sufficient. This paper addresses this question by examining the funreasons that the abstraction works for existing devices, and by showing that these reasons also hold for MEMStores. This result is shown to hold through several case studies of proposed roles MEMStores may take in future systems and potential policies that may be used to tailor systems' access to MEMStores. With one noted n, today's storage interfaces and abstractions are as suitable for MEMStores as for disks.
Steven W. Schlosser, Gregory R. Ganger