A novel workflow based data processing platform as a service

Abstract

This paper presents a novel platform as a workflow engine service which helps users to utilize cloud computing for addressing problems of large scale data processing without knowledge of underlying configuration details for computing applications. Our platform provides elastic resources for workflow applications to ensure that users leverage resources efficiently with the consideration of their budgets. To achieve this goal, we propose a cost-oriented scheduling strategy to guarantee workflow applications to match appropriate resources and to be scheduled in a minimum makespan, These workflows consist of serial and parallel tasks. In addition, we discuss the architecture and the key modules of our platform.