Energy-Efficient Algorithm for Load Balancing and VMs Reassignment in Data Centers

Abstract

Nowadays, cloud computing is an effective solution for providing computing services to consumers. However, data centers that host computing resources are still faced with a misuse of resources and a workload imbalance of servers, where some servers become overloaded while others are underloaded or even idle. This results in performance degradation and resource wastage. The load balancing is a key aspect and has an important role in the management of cloud data centers. It allows an optimal use of the resources and improves the desired Quality of Service (QoS) using optimal methods for allocating resources and distributing workload.In this paper, we propose a load-balancing algorithm that is based on a new parameter called the balance factor of the data center, introduced here, to determine if a data center is imbalanced or not, in order to redistribute the workload equally over all the hosts. To minimize the energy consumption of the data center, our strategy relies on the live migration of virtual machines (VMs) while using a mathematical model. The simulation results, using the CloudSim toolkit, have shown that the energy efficiency can be managed by reassigning VMs to the data-center hosts.

Latency and network aware placement for cloud-native 5G/6G services

“…The VNF placement problem is a long-standing research topic [6] that is usually framed as an optimization problem considering mostly the resource utilization (computing, storage or network capacity) [2], [7], [3] and to a lesser extent the latency, which is the focus of our work. In order to reduce the latency, Alleg et al [8] rely on Mixed-Integer Quadratically Constrained program to find the optimal placement of a chain of network Functions, which is the one that minimizes the processing delay.…”

“…The placement of Virtualized Network Functions (VNFs) has been widely studied in the literature. Most research works focused on load balancing, examine the resource (CPU, RAM, disk) needs of network functions with respect to the resource availability in data centers [2], [3]. Only a few works on microservice placement address latency requirements by considering either the processing delay or the link capacity.…”