On Optimal Heterogeneous Regenerating Codes

TL;DR

This paper introduces a new family of graph-theoretic codes designed for heterogeneous distributed storage systems, achieving optimal file size bounds despite nodes having different storage capacities and repair bandwidths.

Contribution

It presents a novel construction of regenerating codes tailored for heterogeneous DSS, achieving the fundamental bounds on file size.

Findings

Codes achieve the fundamental bound on file size.

Construction is based on graph theory.

Applicable to systems with diverse node capacities.

Abstract

Heterogeneous Distributed Storage Systems (DSSs) are close to the real world applications for data storage. Each node of the considered DSS, may store different number of packets and each having different repair bandwidth with uniform repair traffic. For such heterogeneous DSS, a failed node can be repaired with the help of some specific nodes. In this work, a family of codes based on graph theory, is constructed which achieves the fundamental bound on file size for the particular heterogeneous DSS.