Resource Minimization for Real-Time Applications Using Computer Clouds

In this paper, we address the resource minimization problem for DAG-based real-time applications using computer clouds to: (1) guarantee the satisfaction of a real-time application's end-to-end deadline, (2) ensure the number of computers allocated to the application is minimized, and (3) under allocated resources, minimize the application's make span. We first give lower and upper bounds for resources needed to guarantee the satisfaction of a real-time application's deadline. Based on the bounds, we develop a heuristic algorithm called minimal slack time and minimal distance (MSMD) algorithm that finds the minimum number of computers needed to guarantee the application's deadline and schedules tasks on the allocated resources so that the application's make span is minimized. Our experimental results show that the MSMD algorithm can guarantee applications' end-to-end deadlines with less resources compared with other heuristic scheduling algorithms existed in the literature. In addition, under the minimal allocated resources, the MSMD algorithm can, on average, reduce an application's make span by 10% of its deadline.