[Paper Review] A Randomly Expandable Method for Data Layout of RAID Storage Systems

De-zhai Yuan, Xing-yi Peng+2 more|arXiv (Cornell University)|Sep 14, 2017

Advanced Data Storage Technologies25 references3 citations

TL;DR

This paper proposes a novel RAID data layout scheme using Random Binary Extensive Code (RBEC), a systematic XOR-based code that enables fully random expandability of data and parity disks, high fault tolerance, and quasi-optimal storage efficiency. The method supports arbitrary fault tolerance through simple XOR operations and dynamic expansion based on system needs, making it ideal for scalable, reliable storage systems.

ABSTRACT

With the increase of huge amounts of data in volume, velocity, and variety, the need for capacity of Redundant Arrays of Inexpensive Disks (RAID) storage systems is dramatically growing. However, the probability of disk failures in RAID storage systems is sharply high with the increase of program/erase cycles, read cycles, and retention time. Furthermore, they are faced with more challenges in fault tolerance, storage efficiency, computational complexity, and expandability. This article presents a novel data layout scheme for RAID storage System using Random Binary Extensive Code (RBEC), which are designed to ensure random expandability, high reliability, and availability of data in RAID storage systems. RBEC is a family of systematic code, in which the generator matrix consists of two submatrices with entries over GF(2), an identity matrix on the top, and another submatrix on the bottom. Compared with the existed approaches, the attractive advantages of our schemes include 1) they are completely implemented based on only simple eXclusive OR (XOR) operations and have systematic code property, 2) they can provide arbitrary fault tolerance, 3) their storage efficiency is quasi-optimal, and 4) data and parity disks of RAID storage systems can be randomly expanded according to practical requirements. Thus, our scheme is particularly suitable for RAID storage systems that need higher reliability, availability, and expandability.

Motivation & Objective

• To address the growing demand for scalable, reliable, and expandable RAID storage systems under increasing data volume and failure rates.
• To overcome limitations in fault tolerance, storage efficiency, and expandability of existing RAID schemes.
• To design a data layout method that supports arbitrary fault tolerance and dynamic, random expansion of data and parity disks.
• To achieve high reliability and availability using only simple XOR operations, minimizing computational complexity.

Proposed method

• The proposed scheme uses Random Binary Extensive Code (RBEC), a systematic code with a generator matrix composed of an identity matrix and a binary submatrix.
• Data and parity disks are organized such that the generator matrix enables systematic encoding with only XOR operations.
• The method supports arbitrary fault tolerance by allowing flexible code parameters based on system requirements.
• Random expandability is achieved by dynamically adding new data and parity disks without restructuring existing data.
• The design ensures that all operations are performed over GF(2), minimizing computational overhead.
• The scheme maintains quasi-optimal storage efficiency by balancing data and parity disk usage across the array.

Experimental results

Research questions

• RQ1 How can RAID storage systems achieve truly random and dynamic expandability of data and parity disks without data migration or restructuring?
• RQ2 What coding scheme enables high fault tolerance while maintaining low computational complexity using only XOR operations?
• RQ3 Can a systematic code be designed to support arbitrary fault tolerance levels in RAID systems with minimal storage overhead?
• RQ4 How does the proposed scheme compare to existing RAID codes in terms of storage efficiency and expandability?
• RQ5 What is the impact of using a binary submatrix in the generator matrix on fault tolerance and system scalability?

Key findings

• The RBEC-based data layout scheme supports arbitrary fault tolerance levels through flexible code parameterization.
• The scheme achieves quasi-optimal storage efficiency, approaching the theoretical lower bound for redundancy in RAID systems.
• All encoding and decoding operations are implemented using only XOR operations, significantly reducing computational complexity.
• The method enables completely random expansion of both data and parity disks, allowing dynamic scaling based on real-time storage demands.
• The systematic nature of RBEC ensures that data is stored in original form, simplifying access and improving performance.
• The approach maintains high reliability and availability by combining strong fault tolerance with efficient, low-overhead operations.