Reliability Model of Fault-Tolerant Dual-Disk Redundant Array

Abstract

This paper deals with data storage systems based on fault-tolerant dual-disk RAID-1, which are widely used as high-reliable data storage systems and have acceptable overhead expenses in hardware implementation. Advanced reliability model of dual-disk RAID-1 array based on Markov chains, which takes into consideration finite time of disk replacement after disk failure, different disk failure rate in array’s normal and rebuild states, and probability of read errors during array rebuild procedure, are also overviewed in this paper. Mathematical solution of reliability model, calculation formula for mean time to data loss, estimation of disk and array reliability parameters and MTTDL calculation example are also provided.

Reliability model of fault-tolerant data processing system with primary and backup nodes

“…As for failure and recovery we assume that both nodes are independent. The advanced model is a further development of reliability models obtained by authors earlier [5]. Now let us introduce the following state-space for the system with primary and backup nodes, and conditions for transition from one state to another according to the advanced reliability model:…”

“…On the other hand, a number of specialized books [3,4], dedicated to reliability of computing systems and networks, discuss data processing systems, but the given reliability models for duplex systems are too simplified and provide overestimated values for reliability indexes. Within the scope of the scientific research work of authors regarding reliability of data processing and transmission systems [5] a scientific task of development of a specialized reliability model for the dual-node data processing system with primary and backup nodes has been raised. This model is also considered for future application to obtained results in designing of data processing systems for industrial enterprises.…”

Using a specialized Markov chain in the reliability model of disk arrays RAID-10 with data mirroring and striping

“…Over the past few years, within the ambit of the research work in the field of reliability models of data transmission, processing and storage systems [9,10], the author developed a specialized Markov Chain for the reliability model of the fault-tolerant system with n elements, dual-type failures, nonsynchronized repairs and full restore after reaching the failed state, and obtained a generalized procedure for calculation of the mean time to failure and a stationary availability factor.…”