Contribuciones a la optimización de consumo energético en centros de procesamiento de datos
Document typeBachelor thesis
Rights accessOpen Access
There is currently a problem not only environmental but also at the economic level in the industry of datacenters, since it is estimated that CO2 emissions has reached 2% worldwide. There are different projects, theories and research on how to reduce consumption in order to minimize their economic and environmental impact. This project focuses on analyzing the problem, identify the various factors consumer choose virtualization as a factor to solve, analyze the state of the art, propose a solution and develop a system able to implement the solutions following the premise of not modifying the software control and virtualization that is taken as basis for development. A scenario has been created in order to mimic a data center. The OpenStack platform has been used as the basis of control data center. At the basic installation, has been added the telemetry module to monitor different resources. There has been some changes to the configuration in order to achieve our goals. Four algorithms have been developed as a basis for resource optimization. The algorithms have been created to be scalable and able to accommodate future modifications and improvements. We have developed a REST API capable of applying the different solutions. Having a library capable of communicate our API with OpenStack has helped and reduced our time in developing the software. However, there have been important changes in the pkgcloud library in order to achieve all the functionality required by our API. We have managed to make a robust system, able to operate autonomously and with the possibility of adding new features and enhancements. Our system has been tested in critical cases obtaining good results.