Uma estratégia baseada em modelos e algoritmos genéticos para otimizar arquiteturas de data centers

Abstract

A growing concern in society is related about sustainability and the environmental impacts caused by energy consumption and generation. In deed, there is great global pressure for companies to adopt sustainable practices, not only because of the nancial savings from reducing energy consumption, but also due to the awareness of the depletion of natural resources for future generations. Considering the advancements of technology, the society needs more interconnected services to the Internet and, consequently, the infrastructure demanded to support for these services also contributes with a signi cant impact on the consumption of electricity. In recent years, due to factors such as social networking, cloud computing, and ecommerce, data center has been growing in importance and, so, its reliability, collaboration, availability, and cost reduction represent elements under studies. The data center systems require an entire infrastructure with redundancy mechanisms to operate with high availability, a fact that may have a negative impact on both the electrical energy consumption and operational cost. This work proposes a method based on multiobjective genetic algorithms to optimize cost, environmental impact and availability of electrical energy infrastructures of data centers. The main goal is to maximize the availability, minimize the total cost and the operational exergy (used to estimate the environmental impact). To compute such metrics, models were proposed using the Energy Flow Model (EFM), Reliability Block Diagram (RBD) and Stochastic Petri Net (SPN). Two case studies are conducted to show the applicability of the proposed strategy: (i) takes into account 5 typical data center architectures that were optimized to conduct the validation process of the proposed strategy; (ii) uses the optimization strategy in four architectures classi ed by ANSI / TIA-942 (TIER I to TIER IV). In both case studies, signi cant improvements were achieved in the results, which were very close to the optimum one that was obtained by a brute force algorithm that analyzes all the possibilities and returns the optimal solution. It is worth mentioning that the time used to obtain the results using the genetic algorithm approach was signi cantly lower (6,763,260 times), in comparison with the strategy which combines all the possible combinations to obtain the optimal result.