Simulation-based Cost Analysis for Distributed Systems

Abstract

Since 2006 more enterprises are migrating their application stack from an on-premises to a cloud infrastructure. An important difference between both infrastructures is the costing structure. The operational cost of a cloud deployment is more difficult to determine than for on-premises deployments. This is, because of the complex price structure of cloud deployments and the dynamic use of resources. The result of this is that enterprises cannot estimate the operational costs of their application stack when they are moving them to a cloud infrastructure. This is not preferable, because the migration to the cloud is mainly based on financial motivation. The goal of this study is to create a framework which assists enterprises, in estimating the operational cost of their application stack when migrated to the cloud. A tool including a graphical editor and simulation engine is created, guiding software architects to model a software system’s architecture and to analyse that, based on operational cost. The user experience of our framework is evaluated by a case study with software architects and the results of the simulation are verified in a series of experiments. The outcomes are promising, and certain interesting opportunities are identified for future research.