ABSTRACT
This chapter describes the architecture of the example cloud-redundant arrays of independent disk (RAID) 5 systems and its reliability modelling using multi-state fault trees. It presents methods for evaluating the disk-level and system-level state probabilities of the example cloud-RAID 5 system. Effects of functional dependence behavior between the RAID controller and disks are addressed. Numerical results demonstrate effects of each coverage factor on the disk or system state probabilities. Moreover, these results show that failure to consider imperfect fault coverage leads to inaccurate disk or system state probabilities, thus misleading system design, maintenance, and optimization activities. The chapter presents results of system state probabilities under three cases: the controller is perfectly reliable, has perfect fault coverage, and has imperfect fault coverage. While the cloud-RAID 5 system is used as an illustrating case study, the proposed methodology is applicable to other levels of cloud-RAID storage systems such as cloud-RAID 6.
