Ranking of Cloud Service Providers Using a Dynamic TOPSIS Model for Provisioning of Enterprise IT Infrastructure in the Cloud Open Access
Downloadable ContentDownload PDF
Cloud computing has disruptively created an opportunity for businesses to effectively use their Internet Technology (IT) infrastructures. Consequently, it is increasingly becoming an IT solution for enterprises that are hard-pressed to operate within an efficient, innovative, cost effective and flexible environment. That being the case, provisioning enterprise IT infrastructure in the cloud is a rather daunting task. A rising number of government agencies and private enterprises turning to the cloud for shared access, sufficient data storage and scalable computing resources has flooded the cloud computing marketplace with myriad cloud-based services, implementation options and Cloud Service Providers (CSPs). This has compounded decision makers’ confusion and uncertainty about the best-suited IT solution for their business environment. Multi-Criteria Decision Making (MCDM) techniques are commonly used by decision makers to rank CSPs and to select the most suitable cloud solution for their business needs. Among the most commonly used MCDM techniques is the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS). Like other MCDM techniques, however, TOPSIS has its own limitations. One critical limitation of TOPSIS is the lack of proportional weight adjustment for local criteria scores. That is, if criteria scores in close proximity to one another are mapped to single values (e.g., average values) rather than score ranges as expected or required by decision makers and if no other mechanism is implemented to proportionally adjust criteria weights for their corresponding score ranges, then the classical TOPSIS techniques will result in an under-provisioning of cloud resources or selecting suboptimal CSP. In this paper, a new, simple, flexible and robust MCDM model dubbed Dynamic TOPSIS Model (DTM) is proposed as an enhancement to the classical TOPSIS model. DTM addresses the weight adjustment problem by assigning criteria weights for a range of values, rather than for just single values. DTM also will allow decision makers to adjust score thresholds based on business requirements and system constraints. To demonstrate the applicability of DTM in selecting suitable CSPs for provisioning of enterprise IT infrastructure in the cloud, a case study is presented at the end of this paper that uses real-world cloud data for ranking 39 cloud services from 11 CSPs based on Quality of Service (QoS) attributes.