Reliability Analysis And Optimization Of Cloud Data Storage Multi-state Systems Considering Imperfect Coverage

Rapidly growing data requires more reliable data storage methods.Cloud data storage is one of the key technologies for data storage,that use multiple redundancy techniques to achieve high data reliability and availability.Imperfect coverage as an inherent behavior of any fault-tolerant system,may cause a system failure due to the imperfect fault recovery mechanism,even if adequate redundancy still remains.The imperfect coverage can lead to wasted redundancy and unreliable systems.In view of this,this paper is aimed at the reliability analysis and optimization of the 5-disk cloud storage system affected by the imperfect coverage.Two types of imperfect coverage are considered,including the Element Level Coverage(ELC)and the Fault level Coverage(FLC).Specific research work and contributions are reflected as follows:(1)The reliability analysis method of cloud data storage multi-state system influenced by Element Level Coverage.Considering the feature of Element Level Coverage,the failure coverage factors of the disks in the system are independent and do not affect each other.Analyzing on the disk failure and recovery mechanism,it is concluded that a disk can assume four states and the states and state-transitions of the multi-state disk under the influence of ELC are represented using the Markov state transition model.Based on the Laplace transform,the calculation formula of the disk state probability are derived.Based on the disk state probabilities calculated,the state probabilities of the entire 5-disk cloud data storage system are analyzed using a multivalue multi-state decision diagram.In addition,numerical simulations are performed to verify the correctness of the method.The effects of state transition rates and mission time on the state probabilities of the disk are studied.The influence of the disk coverage factor on the cloud system state probabilities is also studied.(2)The reliability analysis method of cloud data storage multi-state system influenced by Fault Level Coverage.The main feature of the fault level coverage if the system's fault coverage capability is dependent on the number of disk faults happening in a certain time window.This work considers the reliability of the cloud data storage system in two cases: the controller runs perfectly and the controller may undergo failures.The effects of disk f coverage factors,controller failure parameters and mission time on the overall system state probabilities are studied.(3)This study considers the reliability optimization(i.e.,the disk supplier selection problem)of cloud data storage systems under the influence of the imperfect coverage.Three optimization problems are proposed: maximizing system reliability without constraints,maximizing system reliability under a certain cost limit and minimizing system cost under a certain reliability limit.The proposed problems are solved for a 5-disk cloud storage system,and the results show that solving the proposed optimization problems can make good choices of disk supplier combination leading to reliable and cost effective design of the cloud storage system.