Ensuring Data Storage System Fault Tolerance Using Redundant Modular Codes

Abstract

To ensure the reliability of information storage in server and other data storage systems, fault tolerance technologies are widely used. RAID technologies are the most popular among them, as they use various methods to prevent information loss. These methods include duplication, parity control, and the use of Reed-Solomon error correction codes. Additionally, they combine several methods to provide maximum fault tolerance. Redundancy in information presentation and hardware costs for implementing such systems range from 25% to 100% and even 200% in most cases. In this article, we propose using redundant modular codes to create reliable data storage systems. These codes represent information as the remains of a division into coprime modules, allowing for parallel storage of data on multiple disks. This results in a data storage system that can correct errors on the fly when individual disks fail, as well as provide the ability to hot-swap faulty disks. Additionally, the proposed system has lower hardware costs than RAID technologies.