Fault Tolerance for Real Time Application in Cloud Computing

Abstract

Cloud computing is a web based technology which is the next step in the evolution of distributed computing. It provides various benefits which can be provisioned easily on-demand. Real time systems can take advantages of cloud computing infrastructure to executes its tasks. In real time systems, every real time tasks have a fixed time limits before which they must complete their execution. Sometime failure of real time tasks can be catastrophic. To ensure fault free and successful execution of real time tasks we need a robust fault tolerance technique. Task replication is one of the technique in cloud computing to tolerate the faults but it incurs resource wastage due to executing same task several times on different virtual machines. In this work, we have presented a novel fault tolerance model which uses task replication technique to tolerate the faults for real time applications in cloud computing and also optimize the number of replica for each tasks to reduce the resource wastage. In our proposed model, each task is assigned to a number of virtual machines which have high reliability value using FT load Distributor and Replica Predictor module. Reliability value of each virtual machine gets updated after every successful or unsuccessful execution of task by the virtual machine. Each replica is independently executed on separate Virtual Machine and the result is sent to adjudicator node which selects the final output replica.

We have simulated our proposed model using cloudsim 3.0.3 simulator and have found that for high load our model gives approx. 85% success,turnaround time is approximately 30.63% less and better throughput than other fault tolerance models. Also, we can save approx. 37% of computing resources.