A Novel Hierarchical Scheduling Method for Managing Parallel Workloads in Cloud

Nowadays one of the most useful computing paradigm is cloud computing which can be accessed as “Pay-n-Use” method and it provides services in terms of software, utility and platform. Cloud computing leads people to run any kind of complex applications and it allows accessing data centers remotely and costing effectively. It is well known that most of the complex applications need parallel processing efficiency. Transferring data, messages, any kind of communication and synchronization are some of the parallel processes involved in the cloud where these processes don’t require CPU utilization high due to parallelism. Also cloud computing supports parallel processing, distributed processing and basically it is multi-tenant environment. Scheduling the task for parallel processing among multi-tenant involves time complexity and it affects the performance of the cloud computing. The existing studies were concentrating on consolidating the parallel workloads using virtualization techniques. But in this paper, a novel hierarchical scheduling approach is used to schedule the workloads by considering the attributes based weight value, generated dynamically. The weight value is calculated by examining the various attributes of the workloads in terms of resources, time taken to complete the job and availability of the resources. The best value of the attributes is elected by Artificial Immune System and the server mode is verified if it is in Sleep, Idle or in Busy and assign the job in order to save the energy. The simulation based experiment is carried out in GreenCloud software and the performance is evaluated.