The large success of the Cloud computing, its strong impact on the ICT world and on everyday life testifies the maturity and effectiveness this paradigm achieved in the last few years. Presently, the Cloud market offers a multitude of heterogeneous solutions. However, despite the undeniable advantages, Cloud computing introduced new issues and challenges. In particular, the heterogeneity of the available Cloud services and their pricing models makes the identification of a configuration that minimizes the operating costs of a Cloud application, guaranteeing at the same time the Quality of Service, a challenging task. This situation requires new processes and models to design software architectures and predict costs and performance considering together the large variability in price models and the intrinsic dynamism and multi-tenancy of the Cloud environments. This work aims at providing a novel mathematical approach to this problem presenting a queuing theory based Mixed Integer Linear Program (MILP) to find a promising multi-cloud configuration for a given software architecture. The effectiveness of the proposed model has been favorably evaluated against first principle heuristics currently adopted by practitioners. Furthermore, the configuration returned by the model has been also used as initial solution for a local-search based optimization engine, which exploits more accurate but time-consuming performance models. This combined approach has been shown to improve the quality of the returned solutions by 37% on average and reducing the overall search time by 50% with respect to state-of-the-art heuristics based on tiers utilization thresholds.
Ciavotta, M., Ardagna, D., Gibilisco, G. (2017). A mixed integer linear programming optimization approach for multi-cloud capacity allocation. THE JOURNAL OF SYSTEMS AND SOFTWARE, 123, 64-78 [10.1016/j.jss.2016.10.001].
A mixed integer linear programming optimization approach for multi-cloud capacity allocation
Ciavotta, M
;
2017
Abstract
The large success of the Cloud computing, its strong impact on the ICT world and on everyday life testifies the maturity and effectiveness this paradigm achieved in the last few years. Presently, the Cloud market offers a multitude of heterogeneous solutions. However, despite the undeniable advantages, Cloud computing introduced new issues and challenges. In particular, the heterogeneity of the available Cloud services and their pricing models makes the identification of a configuration that minimizes the operating costs of a Cloud application, guaranteeing at the same time the Quality of Service, a challenging task. This situation requires new processes and models to design software architectures and predict costs and performance considering together the large variability in price models and the intrinsic dynamism and multi-tenancy of the Cloud environments. This work aims at providing a novel mathematical approach to this problem presenting a queuing theory based Mixed Integer Linear Program (MILP) to find a promising multi-cloud configuration for a given software architecture. The effectiveness of the proposed model has been favorably evaluated against first principle heuristics currently adopted by practitioners. Furthermore, the configuration returned by the model has been also used as initial solution for a local-search based optimization engine, which exploits more accurate but time-consuming performance models. This combined approach has been shown to improve the quality of the returned solutions by 37% on average and reducing the overall search time by 50% with respect to state-of-the-art heuristics based on tiers utilization thresholds.File | Dimensione | Formato | |
---|---|---|---|
article.pdf
accesso aperto
Descrizione: Pre-Print A Mixed Integer Linear Programming Optimization Approach for Multi-Cloud Capacity Allocation
Tipologia di allegato:
Submitted Version (Pre-print)
Dimensione
2.3 MB
Formato
Adobe PDF
|
2.3 MB | Adobe PDF | Visualizza/Apri |
1-s2.0-S0164121216301996-main (1).pdf
Solo gestori archivio
Tipologia di allegato:
Publisher’s Version (Version of Record, VoR)
Dimensione
1.73 MB
Formato
Adobe PDF
|
1.73 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.