Exposing non-volatile memory cache for adaptive storage access

Abstract

This paper proposes a method that combines next generation non-volatile (NV) memory technologies to block storage and makes use of NV memory as storage cache. The existing method to combine cache storage with block storage hides the cache storage under the block storage interface. Such a method cannot efficiently utilize the memory access interface and short latency provided by NV memory. The proposed method, which is called VEMS (Virtually Extended Memory Storage), utilizes NV memory as its primary interface and also provides the capacity of the combined block storage. VEMS exposes NV memory rather than hiding it in order to overcome the problems of the existing method and to achieve higher performance. VEMS also introduces adaptive storage access to mitigate the access latency of block storage while providing the memory access interface of NV memory. We implemented VEMS in the Linux kernel. The experiment results showed that VEMS performs considerably better than the existing method to combine cache storage with block storage, and performs comparably with a ramdisk if data sizes fit in the size of NV memory.