Multi-objective optimization for VM placement in homogeneous and heterogeneous cloud service provider data centers
Autor: | Taoufik Aguili, Zacharie Ales, Rym Regaieg, Mohamed Koubaa |
---|---|
Přispěvatelé: | National Engineering School of Tunis = Ecole Nationale d'Ingénieurs de Tunis [University of Tunis El Manar] (ENIT), University of Tunis El Manar, CEDRIC. Optimisation Combinatoire (CEDRIC - OC), Centre d'études et de recherche en informatique et communications (CEDRIC), Ecole Nationale Supérieure d'Informatique pour l'Industrie et l'Entreprise (ENSIIE)-Conservatoire National des Arts et Métiers [CNAM] (CNAM)-Ecole Nationale Supérieure d'Informatique pour l'Industrie et l'Entreprise (ENSIIE)-Conservatoire National des Arts et Métiers [CNAM] (CNAM), Optimisation et commande (OC), Unité de Mathématiques Appliquées (UMA), École Nationale Supérieure de Techniques Avancées (ENSTA Paris)-École Nationale Supérieure de Techniques Avancées (ENSTA Paris) |
Jazyk: | angličtina |
Rok vydání: | 2021 |
Předmět: |
Mathematical optimization
Linear programming Computer science 02 engineering and technology computer.software_genre Multi-objective optimization Theoretical Computer Science Reduction (complexity) Set (abstract data type) Resource (project management) [MATH.MATH-CO]Mathematics [math]/Combinatorics [math.CO] 0202 electrical engineering electronic engineering information engineering ComputingMilieux_MISCELLANEOUS Numerical Analysis business.industry 020206 networking & telecommunications Computer Science Applications Computational Mathematics Computational Theory and Mathematics Virtual machine 020201 artificial intelligence & image processing Data center business computer Software Integer (computer science) |
Zdroj: | Computing Computing, Springer Verlag, 2021, 103 (6), pp.1255-1279. ⟨10.1007/s00607-021-00915-z⟩ |
ISSN: | 0010-485X 1436-5057 |
Popis: | We address the virtual machine placement problem that arises in Cloud Service Providers data centers. We purpose, a Multi-Objective Integer Linear Programming model which aims at optimizing simultaneously the number of hosted virtual machines, the resource wastage and the number of active physical machines (PM) in order to minimize power consumption. This new combination of objectives enables to maximize the client satisfaction rate with minimizing the data center (DC) operational costs. We modelize this problem with a multi-objective integer linear program and solve it through two different methods. The first method computes a unique solution for a given preference order over the objectives whereas the second computes a set of non-dominated solutions. Both methods are compared through extensive simulation scenarios. We consider two DC architectures: homogeneous DCs (i.e., a DC with PMs having the same amount of resources) and heterogeneous DCs. We study the impact of each DC configuration on the performances of the solutions. We show that the second method leads to solutions with a reduction of up to 30% over the number of used PMs and that the heterogeneous DCs outperforms the homogeneous one across all objectives. |
Databáze: | OpenAIRE |
Externí odkaz: |