Non-homogeneous distributed storage systems

TL;DR

This paper introduces non-homogeneous distributed storage systems with a super node, proposing three schemes that improve repair efficiency, reduce complexity, and increase data availability compared to traditional homogeneous systems.

Contribution

It presents three novel distributed storage schemes based on MDS and non-MDS codes tailored for non-homogeneous DSS, enhancing repair efficiency and data availability.

Findings

Achieve optimal repair bandwidth of (k+1/2)(M/k) for 1-node failure.

Require only one quarter of the minimum file size for operation.

Increase data availability by 10% over traditional homogeneous DSS.

Abstract

This paper describes a non-homogeneous distributed storage systems (DSS), where there is one super node which has a larger storage size and higher reliability and availability than the other storage nodes. We propose three distributed storage schemes based on (k+2; k) maximum distance separable (MDS) codes and non-MDS codes to show the efficiency of such non-homogeneous DSS in terms of repair efficiency and data availability. Our schemes achieve optimal bandwidth (k+1/2)(M/k) when repairing 1-node failure, but require only one fourth of the minimum required file size and can operate with a smaller field size leading to significant complexity reduction than traditional homogeneous DSS. Moreover, with non-MDS codes, our scheme can achieve an even smaller repair bandwidth of M/2k . Finally, we show that our schemes can increase the data availability by 10% than the traditional homogeneous DSS scheme.