Paper Title: Maximizing Performance/cost of SSD composed of Memory-type and Storage-type SCMs

Paper Abstract: Although storage-type SCM (S-SCM) based SSD has already appeared like 3D-XPoint, memory-type (M-SCM) based SSD has not been realistic yet due to high bit cost. This paper proposes multi-SCM SSD using M-SCM as non-volatile cache of S-SCM based SSD. In this work, the best configuration of multi-SCM SSD is discussed to balance SSD cost and performance. The optimal capacity ratio of M-SCM and S-SCM strongly depends on the application characteristics. In case of the hot and random workload, only 5% capacity of M-SCM improves the overall performance of SSD by up to 134% with 12% cost increase. Additionally, to enhance SSD performance, effective wear-leveling policy is also considered.

Paper Author: Reika Kinoshita, Student, Chuo University/University of Tokyo (Japan)

Author Bio: Reika Kinoshita received the B.E. degree from the Department of Electrical, Electric, and Communication Engineering, Chuo University, Tokyo, Japan, in 2018, where she is currently pursuing the M.S. degree. She has been engaged in a research on performance and reliability of storage-class memory-based SSD.