Effect of Lazy Rebuild on Reliability of Erasure-Coded Storage Systems

Abstract

Erasure-coding redundancy schemes are employed to ensure improved reliability of storage systems against device failures. The effect of the lazy rebuild scheme on the Mean Time to Data Loss (MTTDL) and the Expected Annual Fraction of Data Loss (EAFDL) reliability metrics is evaluated. A theoretical model that considers the effect of latent errors and device failures is developed. Analytical reliability expressions for the symmetric, clustered, and declustered data placement schemes are derived. It is demonstrated that the employment of lazy rebuild results in a reliability degradation of orders of magnitude. Independently of whether lazy rebuild is used, for realistic values of sector error rates, the results obtained demonstrate that MTTDL degrades, whereas EAFDL remains practically unaffected. It is also shown that the declustered data placement scheme offers superior reliability.