Benchmark de Infraestruturas para Suporte a Alta Disponibilidade de Banco de Dados

Abstract

Currently, various kinds of computing applications, which are database intensive, are being disseminated. Often, these applications demand a cluster-based infrastructure in order to provide high availability, scalability and fault tolerance. In this scope, this paper presents a study on the behavior of database intensive applications, using open source software, based on different computational structures. Among the infrastructure implemented, were used as high-performance structures, such as clusters. The structural variability was due to the nodes used as physical machines, virtual machines and a cluster built on a cloud computing, a traditional structure also served as a basis for the experiments. The application running on the structures consists of a PostgreSQL database and the performance was measured by the benchmark tool PgBench. The results were drawn graphics that make it possible to visualize the performance variability obtained on different architectures.