[Paper Review] A Framework of Constructions of Minimum Storage Regenerating Codes with the Optimal Update/Access Property for Distributed Storage Systems Based on Invariant Subspace Technique.

Jie Li, Xiaohu Tang+1 more|arXiv (Cornell University)|Nov 20, 2013

Advanced Data Storage Technologies3 citations

TL;DR

This paper presents a generic framework for constructing systematic minimum storage regenerating (MSR) codes with two parity nodes using invariant subspace techniques. The method enables optimal access and update properties, yielding new codes with the largest known number of systematic nodes under given node storage constraints, while subsuming several existing best-performing codes as special cases.

ABSTRACT

In this paper, we present a generic framework for constructing systematic minimum storage regenerating codes with two parity nodes based on invariant subspace technique. Codes constructed in our framework not only contain some best known codes as special cases, but also include some new codes with good properties such as the optimal access property and the optimal update property. In addition, to the best of our knowledge, two of the new codes have the largest number of systematic nodes with the optimal update property for given store capacity of an individual node.

Motivation & Objective

• To develop a unified framework for constructing systematic minimum storage regenerating (MSR) codes with two parity nodes.
• To achieve optimal update and access properties in distributed storage systems.
• To extend existing constructions by enabling larger numbers of systematic nodes under given node storage capacity.
• To incorporate known high-performing MSR codes as special cases within a single theoretical framework.

Proposed method

• The framework employs invariant subspace techniques to systematically design code constructions for distributed storage systems.
• It ensures optimal update and access properties by leveraging algebraic structures derived from invariant subspaces.
• The method generalizes existing constructions, allowing for the derivation of new codes with improved scalability.
• The framework is built on mathematical properties of subspaces that remain invariant under specific linear transformations.
• It enables explicit construction of codes with two parity nodes while maintaining optimal repair efficiency.
• The approach supports both theoretical analysis and practical implementation of codes with minimal repair bandwidth and access overhead.

Experimental results

Research questions

• RQ1 How can a unified framework be designed to construct systematic MSR codes with optimal update and access properties?
• RQ2 What algebraic structure enables the construction of MSR codes with the largest number of systematic nodes under fixed node storage capacity?
• RQ3 Which known high-performing MSR codes can be derived as special cases within this framework?
• RQ4 How do invariant subspaces contribute to achieving optimal repair efficiency in distributed storage systems?

Key findings

• The proposed framework constructs systematic MSR codes with two parity nodes that achieve optimal update and access properties.
• Several known best-performing MSR codes are shown to be special cases of the proposed framework.
• The framework yields new codes with improved scalability, supporting the largest known number of systematic nodes for a given node storage capacity.
• Two of the newly constructed codes achieve the optimal update property with the highest number of systematic nodes reported to date under the same storage constraints.