Many-Objective Resource Allocation for Elastic Infrastructures in Overbooked Cloud Computing Datacenters Under Uncertainty

Abstract

In cloud computing resource allocation, Virtual Machine Placement (VMP) is one of the most studied problems with several possible formulations and different optimization criteria. The present work summarizes a doctoral dissertation focused on studying Many-Objective Virtual Machine Placement (MaVMP) problems. As first contributions, novel taxonomies were proposed for VMP problems in cloud computing environments, in order to gain a systematic understanding of the existing approaches. Additionally, first formulations of MaVMP problems were proposed in: (1) static MaVMP for initial placement, (2)
semi-dynamic MaVMP with reconguration of VMs and (3) dynamic two-phase MaVMP for complex cloud computing environments under uncertainty. Considering the novelty of the proposed formulations, several methods and algorithms were also proposed to address main identied issues on solving each particular MaVMP problem. Experimental results prove the correctness, effectiveness and scalability of the proposed methods and algorithms in different experimental scenarios even when comparing to state-of-the-art alternatives.