Model-Driven Simulation of Elastic OCCI Cloud Resources. (26th December 2020)

• Record Type: Journal Article
• Authors: Ahmed-Nacer, Mehdi; Kallel, Slim; Zalila, Faiez; Merle, Philippe; Gaaloul, Walid
• Is Part Of: Computer journal. Volume 65:Number 5(2022)
• Page Start: 1144
• Page End: 1166
• Publication Date: 2020-12-26
• Subjects: Amazon Web Services -- cloud computing -- CloudSim -- elasticity -- OCCI -- OCCIware -- pricing -- simulation; Computers -- Periodicals; 005.1
• DOI: 10.1093/comjnl/bxaa159
• Languages: English
• ISSNs: 0010-4620

Abstract

Abstract: Deploying a cloud configuration in a real cloud platform is mostly cost- and time- consuming, as large number of cloud resources have to be rented for the time needed to run the configuration. Thereafter, cloud simulation tools are used as a cheap alternative to test cloud configuration. However, most of the existing cloud simulation tools require extensive technical skills and do not support simulation of any kind of cloud resources. In this context, using a model-driven approach can be helpful as it allows developers to efficiently describe their needs at a high level of abstraction. To do, we propose, in this article, a model-driven engineering approach based on the Open Cloud Computing Interface(OCCI) standard metamodel and CloudSim toolkit. We firstly extend OCCI metamodel for the supporting simulation of any kind of cloud resources. Afterward, to illustrate the extensibility of our approach, we enrich the proposed metamodel by new simulation capabilities. As proof of concept, we study the elasticity and pricing strategies of Amazon Web Services (AWS). This article benefits from OCCIware Studio to design an OCCI simulation extension and to provide a simulation designer for designing cloud configurations to be simulated. We detail the approach process from defining an OCCI simulation extension until the generation and the simulation of the OCCI cloud configurations. Finally, we validate the proposed approach by providing a realistic experimentation to study its usability, the resources coverage rate and the cost. The results are compared with the ones computed from AWS.