A Practical, Integrated Multi-Criteria Decision-Making Scheme for Choosing Cloud Services in Cloud Systems

Abstract

Currently, with the rapid development and broad application of cloud computing technology, companies tend to use cloud services to build their applications or business systems. Selecting a trustworthy cloud service is a challenging multi-criteria decision-making (MCDM) problem. Moreover, decision makers are more inclined to use linguistic descriptions to assess the quality of service (QoS) for cloud services due to the limitation of the decision makers' knowledge and the vagueness of criteria information. Therefore, we propose a practical, integrated MCDM scheme for cloud service evaluation and selection of cloud systems, allowing decision makers to compare cloud services based on QoS criteria. First, to more accurately and effectively express the uncertainty of qualitative concepts, the cloud model is used as a conversion tool for qualitative and quantitative information to quantify linguistic terms. Second, given the shortcomings of traditional differentiating measures between cloud models, a more comprehensive distance measurement algorithm using cloud droplet distribution is proposed for the cloud model. The new distance measurement algorithm is applied to the calculation of cloud model similarity and the gray correlation coefficient. The dynamic expertise weights are determined by calculating the similarity between the expert evaluation cloud model and the arithmetic mean cloud model. Then, we propose a technique for order preference by similarity to an ideal solution (TOPSIS) improved by the grey relational analysis (GRA) to calculate the relative closeness of alternatives to the positive and negative ideal solutions and establish a multi-objective optimization model that maximizes the relative closeness of all alternatives to determine the weights of the criteria. Finally, we reconstructed the QoS evaluation criteria for cloud services from both application and service perspectives, and the classical TOPSIS is applied to generate alternative rankings. The practicability and robustness of the scheme were tested through the cloud service selection problem experienced by a real mining company's scheduling platform, which can provide practical references with the theoretical basis for the selection and evaluation of cloud services.