Skip to main content
Skip main navigation
No Access

Towards optimal thread pool configuration for run-time systems of integration platforms

Published Online:pp 129-147https://doi.org/10.1504/IJCAT.2020.104692

Companies seek technological alternatives to increase competitiveness, an example, are the integration platforms, that develop integration processes in order to connect functionalities and data from applications that compose software ecosystems. Threads are computational resources of the platforms, responsible for integration processes execution. Thus, the configuration of threads has a direct influence on the performance of platforms. However, this is a challenge faced by software engineers, who do this configuration empirically. Our scientific and technical literature review did not identify a systematic approach to find the ideal configuration, which depends on factors such as workload, hardware and integration process. Thus, it is appropriate to seek alternatives for configuration that provide a positive impact on the performance of the run-time system, increase productivity, and reduce costs. Inspired by the Particle Swarm Optimisation meta-heuristic, this article proposes an algorithm that finds the ideal configuration for local thread pool, minimising the total average processing time to improve the execution of integration platforms. The algorithm was implemented and tested using a real-life integration process and its performance measures show the feasibility and efficiency of our proposal, supported by a rigorous statistical analysis of results.

Keywords

enterprise application integration, optimisation, PSO, particle swarm optimisation, meta-heuristics, multi-thread, makespan, workflow, integration patterns