A Load-Based Scheduling to Improve Performance in Cloud Systems

Cloud computing is a type of parallel and distributed system for dynamically provisioning on-demand services. Customers and enterprises can employ cloud services including software, platforms and infrastructure to improve the scalability of their services and handle stochastic demands from users. However, random arrival tasks from users have different load sizes and deadline requirements that need to be satisfied in a cloud system. How to schedule tasks to virtual machines under heavy traffic load still remains a challenge problem for cloud service providers. If high-load tasks occupy shared resources for a long time, it will cause higher waiting times for remaining tasks, which make other tasks fail to be completed within their deadline requirements. In this paper, a load-based task scheduling approach is proposed to improve the system performance when facing heavy-tailed traffic. The main purpose is to schedule arrival tasks effectively and prevent high-load tasks from occupying in the same VM group. A series of experiments and performance analyses are conducted by using the CloudSim simulation tool to evaluate the proposed approach. We also compare the experimental results with other approaches under various situations. Simulation results show that the proposed approach outperforms other approaches in terms of response times and profit values.