Cost-effective and Low-complexity Non-constrained Workflow Scheduling for Cloud Computing Environment

Abstract

Cloud computing possesses the merit of being a faster and cost-effective platform in terms of executing scientific workflow applications. Scientific workflow applications are found in different domains, such as security, astronomy, science, etc. They are represented by complex sizes, which makes them computationally intensive. The main key to the successful execution of scientific workflow applications lies in task resource mapping. However, task-resource mapping in a cloud environment is classified as NP-complete. Finding a good schedule that satisfies users' quality of service requirements is still complicated. Even if different studies have been carried out to propose different algorithms that address this issue, there is still a big room for improvement. Some proposed algorithms focused on optimizing different objectives such as makespan, cost, and energy. Some of those studies fail to produce low-time complexity and low-runtime scientific workflow scheduling algorithms. In this paper, we proposed a non-constrained, low-runtime, and low-time-complexity scientific workflow scheduling algorithm for cost minimization. Since the proposed algorithm is a list scheduling algorithm, its key success is properly selecting computing resources and its operating CPU frequency for each task using the maximum cost difference and minimum cost-execution difference from the mean. Our algorithm achieves almost the same cost reduction results as some of the current states of the arts while it is still low complex and uses less run-time.