Repair for Distributed Storage Systems with Erasure Channels

TL;DR

This paper analyzes the repair process in distributed storage systems over erasure networks, deriving the fundamental tradeoff between storage and bandwidth, and explores methods to improve repair success probability under limited transmissions.

Contribution

It introduces the fundamental storage-bandwidth tradeoff in erasure networks and proposes strategies to enhance repair reliability with limited transmissions.

Findings

Optimal tradeoff bound asymptotically achieved with infinite transmissions

Increasing storage space can improve repair success probability

Connecting more nodes can reduce repair bandwidth but may increase delay

Abstract

We study the repair problem of distributed storage systems in erasure networks where the packets transmitted from surviving nodes to the new node might be lost. The fundamental storage-bandwidth tradeoff is calculated by multicasting analysis in erasure networks. The optimal tradeoff bound can be asymptotically achieved when the number of transmission (packets) goes to infinity. For a limited number of transmission, we study the probability of successful regenerating. Then, we investigate two approaches of increasing the probability of successful regenerating, namely, by connecting more surviving nodes or by increasing the storage space of nodes. Using more nodes may pose larger delay and in certain situation it might not be possible to connect to more nodes too. We show that in addition to reducing repair bandwidth, increasing storage space can also increase reliability for repair.