Enhancing reliability of cloud system through proactive identification of under performing components

Abstract

Services involving cloud computing are advertised by the providers such that they are available all the time; but many of the recent past events have shown how a single or a bunch of component seizures can lead to a system failure affecting the business continuity of consumers. Though reliability of a cloud system depends heavily on the capability of the participating components as a whole, but there needs to be some methodology to monitor and make the whole system work efficiently upto the maximum capability of each respective individual components. In this paper, a mechanism for quantifying the reliability of cloud components is proposed so as to provide reliability as a service. Reliability of cloud is dependent on the nature and status of its individual components. Each component's credibility is evaluated based on the reliability score it obtains while executing its assigned jobs. From these scores faulty or error prone components are identified proactively and the non performing components are given a second chance to revive itself, failing which the defective components undergo further checks of the component's importance, history of past failures and other quality factors. If after these checks it is seen that the component is safe to remove and the system will not be affected in anyway by the removal, then the component is quarantined otherwise the defective component is identified and handled suitably. This method is aimed at giving a highly reliable and non-intermittent cloud service to the consumers not at the expense of getting rid of critical components. � 2016 IEEE.