Análisis de los parámetros de operación del clúster de servidores en el centro de datos de la carrera de Ingeniería en Ciencias de la Computación en la Universidad Politécnica Salesiana Sede Quito Campus Sur

Abstract

In the present technical project are the updates of the values of the performance of the processing infrastructure in flops, due to the increase of resources in the servers, and the calculation of the storage performance in Iops of the Data Center of the career of Engineering in Computer Science from the “Universidad Politécnica Salesiana”. Through benchmarking techniques, an analysis between programs was achieved to define the most appropriate and aligned to the characteristics that were required to measure the performance of processing and storage. Because the Data Center is virtualized, six virtual machines were implemented, one for each server, with the chosen programs. 15-day tests were conducted to determine the specific value of each yield. After capturing the values returned by the tools, specific formulas were used to discover the performance infrastructure result, with the Cluster resulting in a value of 69.02 TFlops and the two independent servers with a value of 7.7627 TFlops. Thus, a final result of 76.85 TFlops. Regarding storage, the final result was 93,580 Iops, of which 52,940 I/O Iops are functional. For this reason, it is defined that there was a notable increase for the processing infrastructure according to the results previously stated in the year 2020, where it was determined that there was a value of 5.60 TFlops in the cluster, and 7TFlops for the only non-virtualized server.