Score: 0

Rack-Aware Minimum Storage Partially Cooperative Regenerating Codes with Small Sub-Packetization

Published: April 25, 2025 | arXiv ID: 2504.18335v1

By: Hengming Zhao, Dianhua Wu, Minquan Cheng

Potential Business Impact:

Fixes broken computer parts using less data.

Business Areas:
Self-Storage Real Estate

In the rack-aware model, there are $\bar{n}$ racks each of which has $u$ nodes with the same storage capacity. Assume that there are $h$ failed nodes uniformly distributed in $\bar{h}$ host racks ( defined as racks containing failed nodes), each rack containing $h/\bar{h}$ failed nodes where $h$ is divisible by $\bar{h}$. Then together with its internal helper nodes, each host rack downloads recovery data from $\bar{d}$ helper racks and repairs its failed nodes. The repair bandwidth is defined as the total inter-rack data transfer required for failures recovery, as the intra-rack communication does not contribute to this cost. The full cooperative repair model requires that each host rack must exchange the data with all the other $\bar{h}$ host racks during the cooperative repair phase. However, in the partial cooperative repair model, each host rack only needs to exchange data with $\bar{h}-\delta\ (1\leq\delta\leq\bar{h}-1)$ other host racks, during the cooperative repair phase. In this paper, we focus on the rack-aware minimum storage partially cooperative regenerating (MSPCR) codes for repairing the $h$ node failures. We first derive the lower bound on the repair bandwidth for rack-aware MSPCR codes using extremal combinatorics, and then construct two classes of optimal repair schemes for rack-aware MSPCR codes with small sub-packetization level. In particular, when $\delta=1$, our second codes reduce to rack-aware minimum-storage cooperative regenerating (MSCR) codes, while achieving an $(\bar{h}+1)$-fold reduction in sub-packetization level compared to known rack-aware MSCR codes.

Country of Origin
🇨🇳 China

Page Count
20 pages

Category
Computer Science:
Information Theory