Multilevel Fault-Tolerance Aware Scheduling Technique in Cloud Environment
K Devi, D. Paulraj
Abstract
In cloud computing, the resources are delivered to the users on demand at a considerable cost. Due to low maintenance and high scalability services, enterprises wish to deploy their newly developed application towards the computing environment. For large scale applications, fault tolerance is an essential task that guarantees the reliability and availability of computing services. In this paper, a multi-level fault tolerance scheduling mechanism is proposed that overcomes the real-time failure in the system. In the first phase, non-functional testing and decision making algorithm is used to find the reliability of virtual machines. Here, the reliability criterion is achieved by Reliable Decision K-Nearest Neighbor (RDK-NN) algorithm that considers only the best reliable virtual machine. In the second phase, high availability is achieved using a scheduling algorithm. For this purpose, a Teaching-Learning Based Optimization (TLBO) scheduling is proposed that provides a better-scheduled set of tasks for the corresponding users. The evaluation of the proposed approach is carried out under Cloudsim platform. The performance is determined in terms of makespan time, failure ratio, performance improvement rate, response time and rejection ratio to estimate the scheduling task. The result shows that the system achieves high reliability and availability of data with a multi-level format in the cloud environment.