An efficient deadline-constrained and data locality aware dynamic scheduling framework for multi-tenancy clouds

This research investigates scheduling and resource allocation strategies for multi-tenant cloud applications and their support platforms. This novel research, from both cloud providers’ and tenants’ perspectives, investigates key issues on how to provide efficient scheduling policies while also meeting QoS requirements, improving system’s throughput and resource utilisation, achieving better data locality, and reducing job completion time. This work proposes a novel fairer resource allocation, a data locality aware scheduler and a deadline constrained job scheduler to meet jobs’ deadlines, and finally proposes a scheduling framework, which comprises each of the above individual plug-in schedulers used in tandem.