RETRACTED: Dynamic load balancing upon the replication and deletion of cloud services

Abstract

In the last decade, the considerable increase of the cloud services use has led to the need to have search and selection techniques that match both the requirements of end users and those of the system. Indeed, to select a cloud service that meet the needs of both system and user is a challenge, due to the several conflicting criteria problem for the user on one hand, and for the system, i.e., the load balancing between Virtual Machines (VMs), on the second hand. Therefore, the main challenge, in this context, is how to ensure the user requirements by maintaining the system performance constraint. To deal with this challenge, we present in this paper an approach based on the cloud service replication on one or more VMs when the number of the user requests will be important at a given moment. This allows better load balancing between VMs by distrusting the users’ requests over them. In addition, it allows to select the best cloud service according to the users need. However, the cloud services replication introduces the problem of the storage space saturation. Thus, our second contribution is to select and delete the cloud service replicas without degradation of the load balancing. The two proposed contributions are based on the MCDM techniques in order to select the VMs that can receive the replica of the cloud service and to select those, which their storage space is overloaded in order to delete the replica cloud service. The experimental results, based on Cloudsim simulator, show that our proposal can effectively achieve good performance (load balancing) and improve the response time.

Optimizing Cloud Service Load Balancing Through Heat Conduction Equation Applications

“…As cloud computing technology has proliferated and evolved, the mechanism of load balancing in cloud services has emerged as one of the key technologies to ensure the reliability and efficiency of services [1][2][3]. The core challenge of load balancing lies in the rational allocation of resources to address the constantly changing service requests and system loads [4,5].…”