Performance Evaluation of Heuristics for Cloud Workload Balancing

Abstract

Cloud computing introduces a new level of flexibility and scalability for providers and clients, because it addresses challenges such as rapid change in Information Technology (IT) scenarios and the need to reduce costs and time in infrastructure management. However, to be able to offer quality of service (QoS) guarantees without limiting the number of requests accepted, providers must be able to dynamically and efficiently scale service requests to run on the computational resources available in the data centers. Load balancing is not a trivial task, involving challenges related to service demand, which can shift instantly, to performance modeling, deployment and monitoring of applications in virtualized IT resources. In this way, the aim of this paper is to develop and evaluate the performance of different load balancing heuristics for a cloud environment in order to establish a more efficient mapping between the service requests and the virtual machines that will execute them, and to ensure the quality of service as defined in the service level agreement. By means of experiments, it was verified that the proposed heuristics presented better results when compared with traditional and artificial intelligence heuristics.