Assessing Cloud Infrastructure Costs in Communications-Intensive Applications

Abstract

By deploying cloud infrastructure services, companies strive at achieving faster time to market, improved scalability, lower up-front costs, and lower IT management overhead, among other benefits. However, in longer term, the use of cloud infrastructure may incur significant costs; furthermore, these costs depend both on the required infrastructure resources and on the mode of infrastructure deployment. Therefore, the choice of a particular deployment mode should be justified with a systematic analysis of the associated costs. In this paper, a model for assessing the costs of alternative cloud infrastructure deployment scenarios is introduced. This model decomposes the infrastructure costs into a set of cost factors derived from academic and practitioner literature. A virtual central processing unit (vCPU) is used in the model as a basic system construction element, to which—assuming a virtualised environment—the identified cost factors are assigned. By estimating and aggregating individual cost factors, the total costs associated with a particular deployment scenario can be derived, analysed and compared against the costs of other available deployment options. The cost analysis in this paper focuses on the case of communication-intensive services, where the network data transfer contributes the most to the overall service cost structure. In particular, the in-house, cloud-based and hybrid infrastructure deployment alternatives are analysed. The results of the analysis suggest that the cost of such applications may be minimized by allocating the communication capacity to a single point of service—either in-house or in public cloud—thus benefiting from the economy of scale present in purchasing the capacity.