Efficient Resource Allocation with Score for Reliable Task Scheduling in Cloud Computing Systems

Parallel task execution in a heterogeneous cloud computing system emerges as NP-complete problem and in the literature, many heuristics behave differently when deployed in various environments. Efficient resource allocation improves reliability and leads to the completion of jobs and minimization of delays. In general, static task scheduling algorithms consider the earliest finish time (EFT) of the task to minimize the makespan. In this work, we propose a new heuristic called Efficient Resource Allocation with Score (ERAS) for reliable task scheduling in cloud computing systems, which considers temporal operational availability of Virtual Machines (VM) by considering various types of delays and EFT to assign a normalized score to the processor for scheduling tasks. Here, the allocation of VMs to tasks is based on a score given to each VM, by considering multiple criteria. The results show that ERAS algorithm gives better performance with increased reliability when compared to existing algorithms that consider only EFT for allocation.