Disaster Recovery with Minimum Replica Plan

Abstract

Cloud computing has emerged as a new paradigm for hosting and delivering computing resources over the Internet. The cloud has become a dominant and preferred method to store large amounts of data and enable the sharing of that data among several users. It also enables the use of pay-as-you-go pricing models. Today’s cloud computing environment has required data centers to increase the amount of available storage. There are two main concerns with cloud storage: data reliability and cost of storage. This research proposed a data replication management in multi-cloud approach that determine the number of replicas (which should be less than 3) which reduce the cloud storage consumption while meeting the data reliability requirements. Furthermore, it proposed a preventive approach for data backup and recovery aiming at minimizing the number of replicas and ensure high reliability for data before the disaster. The approach named Preventive Disaster Recovery Plan with Minimum Replica (PDRPMR) which is a cost-effective mechanism to reduce the number of replications in the cloud to be 1 or 2-replicas only without compromising the data reliability. The name PDRPMR originates from its preventive action checking of the availability of replicas and monitoring of denial of service attacks to maintain data reliability. Several experiments have been carried out to demonstrate that PDRPMR reduces the amount of storage space used by one third to two-thirds compared to typical 3-replicas replication strategies, which in turn reduces the cost of storage. These two metrics have been used most frequently in the literature. In this thesis we focused on the critical factors that influence the Disaster Recovery (DR) plan including, minimizing storage cost, reducing Recovery Time Objective (RTO), ensuring high reliability rate and decrease the number of replicas to be less than 3 (typical number of replicas).