Shadow Replication using Dynamic Core Allocation for Application Fault Tolerance in virtual Environment

Abstract— As the interest for cloud computing keeps on increasing, cloud specialist organizations are confronted with the overwhelming test to meet the targeted SLA agreements, as far as dependability and convenient execution is concerned, while accomplishing cost and energy efficiency. This paper proposes Shadow Replication, a novel fault-tolerance mechanism for cloud computing, which flawlessly addresses fault at scale, while limiting energy utilization and lessening its effect on cost. Energy conservation is achieved by creating dynamic cores rather than static cores. Cores are created by the application of cloudlets. In other words proportionate cores are created. Core failure metrics are considered to be memory capacity, energy and power consumption. In case any of the parameter exceeded threshold value, core is supposed to be faulted and progress is maintained within shadow which is maintained 1 per VM. Progress of deteriorated core is shifted to next core within same VM. In case all the cores within the VM deteriorate, VM migration is performed. Results obtained by allocating cores dynamically reduce energy consumption, latency, cost and maximize fault tolerance rate because of reduced VM migration overhead.

Keywords— Fault Tolerance, Energy Efficiency, Shadow Replication, Dynamic cores, Core Failure, Application level Failure.

I. INTRODUCTION

Cloud Computing has been developed as an attractive option for process and information escalated applications, because of its low costs, on demand provisioning of services and diminished cost of keeping up interior IT framework [1].Cloud computing will proceed to develop and sustain in the market because of the immense advantages it offers to established as well as newly set up IT firms. Late reviews anticipate yearly development rate of 17.7 percent by 2016, making cloud computing the quickest developing area in the product business [2]. Along with the benefits of cloud comes the constant burden of meeting elevated requirements of dependability, viability, and expandability [3]. These frameworks ought to regularly meet hard due dates and the inability to meet the due dates could have unfortunate outcomes. To avoid these circumstances, data centres constantly expand by adding more servers, networking devices and storage components so that high rising demands could be met within fraction of seconds, giving clients’ high quality experience and enhancing performance to the fullest. However, as the size of data centres increase, two major issues are faced namely failures and high energy utilization.
Read More