Stars: an integrated environment for assessing availability, cost and energy consumption of systems

Abstract

Sustainability has received increasing attention from the scientific community, with a strong focus on reducing energy consumption and maintaining nonrenewable resources for future generations. In parallel, the expansion of cloud computing, social networking, and e-commerce has increased the demand for data centers. In this context, tools that support the modeling of data center architectures and compute metrics such as availability, cost, and energy consumption are extremely important. This paper proposes a tool named Stars for modeling data center architectures that can compute such metrics. Besides that, non-specialized users do not need to know the formalism adopted by the engine to compute the desired metrics (e.g., RBD, SPN, and EFM). Furthermore, an optimization algorithm, named Genetic Algorithm, was integrated into the tool to maximize the results achieved through a list of components. This algorithm can find a combination of components for a given data center architecture in a very reduced fraction of time compared to the brute force algorithm. Results achieved showed that it was possible to obtain responses in less than 3 seconds with the genetic algorithm compared to the 255 seconds required by the brute force algorithm.