The modeling approach we propose is based on a new meta-model for modeling network infrastructures in virtualized data centers. This meta-model, which we refer to as Descartes Network Infrastructure (DNI) meta-model, is part of our broader work in the context of the Descartes Modeling Language (DML), an architecture-level modeling language for dynamic IT systems and services.
The DNI Meta-model has been designed to support describing the most relevant performance influencing factors that occur in practice while abstracting fine-granular low level protocol details. Instances of the DNI Meta-model (DNI models) are automatically transformed to predictive stochastic models (e.g., product-form queueing networks or stochastic simulation models) by means of model-to-model transformations.
We present a document–called DNI Manual–where we describe the details of the DNI Meta-model. The main purpose of of this document is to present the DNI Meta-Model, its instances, sub-meta-models, and model transformations in more details than in a usual conference or journal paper. We describe the Meta-models and the transformations in details, give examples of systems modeled using the meta-models, and present the transformations in a step by step manner.
We have designed the DNI manual as a living, periodically-updated document that is rich in diagrams with well structured and brief textual descriptions of the models and transformations. The DNI Manual is free to download. The authors are open for discussion, suggestions and constructive critics.