Reserved, on demand or serverless: Model-based simulations for cloud budget planning

Cloud computing providers offer a variety of pricing models, complicating the client decision, as no single model is the cheapest in all scenarios. In addition, small to medium-sized organizations frequently lack personnel that can navigate the intricacies of each pricing model, and as a result, end up opting for a sub-optimal strategy, leading to overpaying for computing resources or not being able to meet performance goals. In this paper, we: (1) present the results of a study that shows that, in Ecuador, a considerable percentage of companies choose conservative pricing strategies, (2) present a case study that shows that the conservative pricing strategy is suboptimal under certain workloads, and (3) propose a set of models, a tool and a process that can be used by tenants to properly plan and budget their cloud computing costs. Our tool is based on M (t)/M/* queuing theory models and is easy to configure and use. Note that, even though we are motivated by our study of adoption of cloud computing technologies in Ecuador, our tool and process are widely applicable and not restricted to the Ecuadorian context.