Fault Tolerance in Cloud: A Brief Survey

Abstract

The Cloud computing environment has seen tremendous growth in terms of enterprises migrating to the cloud. The Cloud computing environment has been characterized by its support to multi-tenancy as well as rapid elasticity. A cloud service provider needs to adhere to service level agreements agreed with its tenants. As cloud infrastructure is composed of heterogeneous, it ought to have faults when different components interact with each other. Fault tolerance is one critical task in order to overcome faults in a timely manner and smooth operations. Through this paper, we present a brief survey of fault tolerance approaches available in the literature for the cloud computing environment. The fault tolerance techniques in literature have been broadly classified into three categories viz. Reactive approaches, proactive approaches, and resilient approaches. Reactive methods try to recover the system after faults. Proactive approaches try to prevent the system from entering in failure state by predicting faults beforehand. Resilient approaches are relatively new in the area of cloud fault tolerance predict faults for dynamically changing environments to prevent failures. Resilient methods are have used machine learning and artificial intelligence as a key factors for predicting faults.