A modified shuffled frog leaping algorithm for scientific workflow scheduling using clustering techniques

The scientific workflows in the field of science like biology and astronomy are essential in facilitating and automating the scientific data of high volumes and their processing especially in a computing structure that is large. Owing to the large need for resources, a public heterogeneous cloud tends to play a major role in the completion of tasks. The traditional researches falling into the scheduling workflows in cloud applications were focusing on the problems that have a quality of service that is not sufficient for the competitive environment that exists today. There are scientific workflows that consist of several granular tasks which are intensive in terms of data. For a computational granularity that is efficient, the task clustering has a major role to play in reducing the length of the schedule and the utilization of resources. The workflow scheduling is a prominent issue in cloud computing, and this makes an attempt to map workflow tasks to VMs on the basis of various functional needs. The very popular approaches to this are either the static or the dynamic scheduling algorithms that have been based on various heuristics like the Opportunistic Load Balancing (OLB). But, in the case of workflow scheduling, this becomes a non-deterministic polynomial-hard optimization and is a challenge to achieve within an optimal schedule. The proposed work is a vertical node partition that makes use the vertical node partition that make use of a heuristic and novel shuffled frog leaping algorithm (SFLA) technique of clustering for optimal scheduling of scientific workflow. The results of the technique have shown that the SFLA proposed along with the method of clustering has achieved better performance (in terms of makespan and utilization of resources) compared to the SFLA and the OLB without clustering.