Decision support for middleware performance benchmarking

Abstract

Along with the rapid development of computing systems, the heterogeneity amongst them also increases. With the usage of middleware technologies, the systems work together in a homogeneous environment and allow to integrate previously independent applications, together with new developments. Along with the growing popularity of middleware systems, the performance and efficiency of their underlying technology becomes crucial for the business. For that reason, applying benchmarks on such environments is highly required. Current technologies and literature focus on building diverse benchmarks in order to test the performance of the underlying middleware. The development of a standard benchmark for middleware is extremely challenging, as one needs to realistically stress the different software capabilities. However, information on how to do so, is generally missing, thus the users need to arbitrarily make crucial design decisions. This Master’s thesis aims at providing the means to ease the decision-making for selecting the appropriate middleware benchmark and enables the user to make crucial design decisions when the creation of a new middleware benchmark is intended. We propose the creation of the first Decision Support System for middleware performance benchmarking, capable of guiding the user through relevant components of a benchmark. The prototype is based on current web technologies using the REST architectural style and it provides a decision support approach for decision makers considering the creation of a new middleware benchmark or selecting the right middleware benchmark of choice. At the end, we validate through use cases to show that the functionalities of the system are accomplished successfully.