R&D: Approach for RAID-6 Scaling Based on D-Code
Scaling scheme for storage system
This is a Press Release edited by StorageNewsletter.com on May 13, 2020 at 2:17 pmIEEE Xplore has published, in 2019 IEEE 5th International Conference on Computer and Communications (ICCC) proceedings, an article written by Peng Jin, Ping Xie, Zhu Yuan, Yu Hu, Yuan Gao, and Jun Ma, Computer College of Qinghai Normal University, Xining, China, 810008.
Abstract: “RAID-6 is widely deployed in large data center because of high data reliability. As the explosive increment of data, RAID-6 storage system needs to increase capacity by adding new disks. To regain load balance among all data disks, RAID-6 scaling requires to migrate some data blocks from old disks to new disks. However, in the traditional RAID-6 scaling schemes, the RR scheme migrates a large number of data blocks, which results in high overhead for RAID-6 scaling. The Semi-RR scheme cannot satisfy the uniform data distribution after multiple scaling. In order to make RAID-6 storage system meet efficient scaling requirement, this paper proposes a new scaling scheme based on D-code for RAID-6 storage system – DEC. This scheme has two advantages: First, the scheme achieves minimal data migration by migrating some data blocks in the old disks to new disks. Second, when DEC performs data migration, data blocks are migrated to appropriate location, which reduces the overhead of parity updates by using optimized parity. Through our real experimental performance analysis, the data migration ratio of DEC is reduced by 73.33% $\sim$ 97.20%, and the migration time is decreased by 75.93% $\sim$ 92.75% compared with the traditional RR scheme.“