Towards a Task and Resource Aware Task Scheduling in Cloud Computing: An Experimental Comparative Evaluation

Abstract

Cloud computing has been considered as one of the large-scale platforms that support various type of services including compute, storage, compute, and analytic to the users and organizations with high agility, scalability, and resiliency intact. The users of the Cloud are increasing at an enormous rate which also resulted in issues related to handling and scheduling the users’ requested workload effectively and efficiently on the available Cloud resources. The aim of the Cloud service providers is to maximize resource utilization and in turn increased revenue generation. In the last few years, Cloud Task scheduling has been considered as an important area of research for the researchers. As different scheduling heuristics are associated with different underlying assumptions; thus, performing a precise comparison cannot be guaranteed. This work empirically compares and provides an insight into the performance of some renown state-of-the-art task scheduling heuristics concerning the Makespan, average resource utilization ratio, Throughput. Those approaches include task-aware, resource-aware, and some hybrid approaches. The experiments were then extended by evaluating the performance using average response time for all the compared approaches. The simulation experiments are conducted by utilizing Heterogeneous Computing Scheduling Problems (HCSP) and Google Cloud Jobs (GOCJ) benchmark datasets using CloudSim a renowned simulation tool for Cloud. Based on the findings of the comparative analysis and results discussion, we have highlighted some important aspects of the underlying approaches and for future work we will propose a task-cum-resource aware task scheduling approach.

Keywords

• Cloud computing
• resource allocation
• task scheduling
• scheduling algorithms
• load balancing
• performance evaluation