Fuzzy based task prioritization and VM migration of deadline constrained tasks in cloud systems

In the past few years Cloud Computing has become one of the prominent areas of research supporting huge number of distributed applications. Cloud computing can be treated as a way of having utility computing which provides large amount of resources with the help of service providers. There exists a Service Level Agreement (SLA) between cloud service provider and service requester. Cloud being a virtualized environment provides Virtual Machines as processing elements. The workflow with number of tasks will be submitted to cloud by end users. These tasks will be run on VMs for completion of given job. For real time tasks meeting of deadlines is the Quality of Service (QOS) requirement. In this paper we have used the fuzzy logic based approach to improvise the performance of dead line constrained tasks. Fuzzy logic has been applied at two levels, first at the level of assigning priorities to jobs and second on deciding the VM migrations based on the overloaded situation. Over loading of VMs is a common situation in cloud as resources are shared among multiple users. With the virtualization technology, hardware resources of a physical machine are shared among various processing elements i.e. VMs. This resource sharing results in uncertain performance of Virtual Machines because of resource interference between VMs. In this paper fuzzy logic based approach was chosen because it provides a way to represent uncertainty in the dynamic environments like cloud using smaller search areas. With the usage of fuzzy logic it is possible to bring in the performance improvement and hence to improve the quality of service (QOS) as a part of Service level agreement.