A Novel Approach of Flexible Sized Super Block for NVMe-based Solid State Drives (SSDs)

Abstract

For flash and next-generation Solid-State Drives (SSDs), NVMe (Non-Volatile Memory Express) is a revolutionary storage access and transport protocol offering the quickest reaction times and maximum throughput for all corporate workloads. Users now demand ever-faster response times from business and consumer apps, even though these apps are becoming more complicated and resource-dependent. This proposed novel approach introduces an approach to mitigate increases in the Program Erase (P/E) cycle count related to a physical block in a storage device by employing superblocks of varying sizes. The memory device within the storage system comprises multiple dies, each divided into physical blocks. A controller identifies and selects specific physical blocks from these dies for superblock formation and optimizes the superblock configuration based on data characteristics. In constructing the superblock, the controller aligns the superblock size with the data characteristics. This alignment minimizes the data relocation within the superblock, thus reducing the impact on the P/E cycle associated with a superblock.