Uma proposta a sustentabilidade de data centers baseada no paradigma DevOps

Abstract

In response to the growing demand for computing services and the increased workload in information technology environments, data centers face significant challenges related to energy efficiency and compliance with global sustainability goals. Green computing, by promoting the efficient use of computing resources, offers an approach to mitigate the environmental impacts associated with these infrastructures. In this context, this dissertation investigated the application of the DevOps paradigm, with an emphasis on infrastructure automation, as a strategy to promote operational sustainability in data centers. Two approaches to performing routine IT administrative tasks were analyzed and compared: manual execution and automated execution, the latter implemented using the Ansible tool, employed as a practical instance of the adopted paradigm. The study comprised the creation of the experimental environment, the implementation of automation, and the evaluation of its operational effectiveness. Electricity consumption was measured using the IMAMS platform, while CO2 emissions were estimated based on the average annual emission factor of 54.5 gCO2/kWh. The results indicated that, across all evaluated tasks, automation reduced total execution time by approximately 23.4% and electricity consumption by approximately 24.3% when compared to manual execution, directly reflecting a reduction in estimated carbon dioxide emissions, which decreased from 22.72 gCO2 in the manual scenario to 17.19 gCO2 in the automated scenario. Based on these findings, it is concluded that the adoption of DevOps practices supported by infrastructure automation contributes measurably to operational efficiency and to the reduction of environmental impacts in data centers. The main contribution of this work is the proposal of a reproducible experimental approach for the integrated evaluation of execution time, energy consumption, and CO2 emissions associated with system administration tasks, providing technical support to both researchers and infrastructure managers in decision-making regarding the adoption of automation as a strategy for sustainability in IT environments.