Regenerating Codes for Errors and Erasures in Distributed Storage

TL;DR

This paper develops explicit regenerating codes that are resilient to errors and erasures in distributed storage, optimizing data reliability and repair efficiency even under malicious attacks.

Contribution

It introduces error- and erasure-resilient regenerating codes, establishing their optimality and providing conditions for resilience in existing code frameworks.

Findings

Resilient codes are optimal in storage and bandwidth.

Capacity bounds are established for adversarial environments.

Conditions for incorporating resilience into existing codes are provided.

Abstract

Regenerating codes are a class of codes proposed for providing reliability of data and efficient repair of failed nodes in distributed storage systems. In this paper, we address the fundamental problem of handling errors and erasures during the data-reconstruction and node-repair operations. We provide explicit regenerating codes that are resilient to errors and erasures, and show that these codes are optimal with respect to storage and bandwidth requirements. As a special case, we also establish the capacity of a class of distributed storage systems in the presence of malicious adversaries. While our code constructions are based on previously constructed Product-Matrix codes, we also provide necessary and sufficient conditions for introducing resilience in any regenerating code.