Mirrored and Hybrid Disk Arrays: Organization, Scheduling, Reliability, and Performance

TL;DR

This paper reviews various disk array organizations like mirroring and hybrid arrays, analyzing their performance, reliability, and optimization techniques such as scheduling and arm placement.

Contribution

It provides a comprehensive analysis of RAID1, RAID1/0, hybrid arrays, and erasure coding, including their performance and reliability characteristics.

Findings

RAID1/0 offers balanced load but reduced bandwidth upon disk failure.

Hybrid arrays have higher reliability but increased update overhead.

Prioritized read processing and optimized arm placement improve response times.

Abstract

Basic mirroring (BM) classified as RAID level 1 replicates data on two disks, thus doubling disk access bandwidth for read requests. RAID1/0 is an array of BM pairs with balanced loads due to striping. When a disk fails the read load on its pair is doubled, which results in halving the maximum attainable bandwidth. We review RAID1 organizations which attain a balanced load upon disk failure, but as shown by reliability analysis tend to be less reliable than RAID1/0. Hybrid disk arrays which store XORed instead of replicated data tend to have a higher reliability than mirrored disks, but incur a higher overhead in updating data. Read request response time can be improved by processing them at a higher priority than writes, since they have a direct effect on application response time. Shortest seek distance and affinity based routing both shorten seek time. Anticipatory arm placement places arms optimally to minimize the seek distance. The analysis of RAID1 in normal, degraded, and rebuild mode is provided to quantify RAID1/0 performance. We compare the reliability of mirrored disk organizations against each other and hybrid disks and erasure coded disk arrays.