A software-defined fusion storage system for PCM and NAND flash

In SSD-based storage systems, persistent data are stored in the NAND Flash and however manipulated in DRAM, causing the decoupled inefficiency. The data being closer to the processors are much easier to be lost due to the volatile property of DRAM, leading to serious data reliability problems. In the meantime, existing SSD technology exploits the properties of NAND Flash and leverages NAND Flash + controller + FTL architecture to improve system's performance. In this black-box-modeled structure, the data semantic information is hard to be transferred by conventional interface. Hence the SSD firmware fails to make full use of the performance potential of SSD in terms of the semantic information of data. Moreover, the host can't obtain physical characteristics and statistical information about SSD, failing to be used by the file system or I/O scheduling algorithm designed for the disks. In order to address these problems, we propose a software defined fusion storage system for PCM and NAND Flash. PCM can be defined as the same level storage or as a buffer of NAND Flash to reduce the WA (Write Amplification) of Flash and improve reliability. In this system, we expose the channels, erases counts and data distribution of PCM/NAND Flash to the host, design FTL algorithm close to file system to obtain more semantic information of data accessing, and manage the storage device as a non-transparent structure. To achieve these design goals, we implement a Host Fusion Storage Layer (HFSL) that supports flexible I/O schedule algorithm and address mapping of variable allocation size in the persistent superior performance of SSD. Extensive experimental results demonstrate the efficiency of the proposed schemes.