Acquisition planning and scheduling of computing resources. (December 2016)

• Record Type: Journal Article
• Authors: Yang, Chien-Nan; Lin, Bertrand M.T.; Hwang, F.J.; Wang, Meng-Chun
• Is Part Of: Computers & operations research. Volume 76(2016)
• Page Start: 167
• Page End: 182
• Publication Date: 2016-12
• Subjects: Computing service -- Acquisition planning -- Scheduling -- Heuristics -- Tabu search -- Genetic algorithm; Operations research -- Periodicals; Electronic digital computers -- Periodicals; 004.05
• DOI: 10.1016/j.cor.2016.06.015
• Languages: English
• ISSNs: 0305-0548

Abstract

Abstract: Cloud computing has been attracting considerable attention since the last decade. This study considers a decision problem formulated from the use of computing services over the Internet. An agent receives orders of computing tasks from his/her clients and on the other hand he/she acquires computing resources from computing service providers to fulfill the requirements of the clients. The processors are bundled as packages according to their speeds and the business strategies of the providers. The packages are rated at a certain pricing scheme to provide flexible purchasing options to the agent. The decision of the agent is to select the packages which can be acquired from the service providers and then schedule the tasks of the clients onto the processors of the acquired packages such that the total cost, including acquisition cost and scheduling cost (total weighted tardiness), is minimized. In this study, we present an integer programming model to formulate the problem and propose several solution methods to produce acquisition and scheduling plans. Ten well-known heuristics of parallel-machine scheduling are adapted to fit into the studied problem so as to provide initial solutions. Tabu search and genetic algorithm are tailored to reflect the problem nature for improving upon the initial solutions. We conduct a series of computational experiments to evaluate the effectiveness and efficiency of all the proposed algorithms. The results of the numerical experiments reveal that the proposed tabu search and genetic algorithm can attain significant improvements. Abstract : Highlights: An acquisition planning model of computing resource is introduced. The model incorporates scheduling decisions. Several solution approaches are proposed. Performances of the approaches are examined through computational experiments.