Dimensioning the relationship between availability and data center energy flow metrics

Abstract

The advancement of technology and the growing number of applications available to network users have increased the demand for services hosted in cloud environments. In 2020, more than 4 billion of people access these services through the Internet, a value 7% higher in comparison to the same period in 2019. To support the demand for such services, an environment that provides such conditions for applications available whenever needed has grown in importance. These environments are generally available from large data centers, which consume large amounts of electricity to provide such demand service capacity. In this context, this work proposes an integrated and dynamic strategy that demonstrates the impact of the availability on the energy consumption of the devices that compose the data center system architecture. In order to accomplish this, colored Petri net models were proposed for quantifying the cost, environmental impact and availability of the electric energy infrastructure ofdata centers. The models presented in this work are supported by the developed prototype. Two case studies illustrate the applicability of the proposed models and strategy. Significant results were obtained, showing an increase close to 100% in the system availability, with practically the same operational cost and environmental impact.