Ranking enterprise application integration platforms from a performance perspective: An experience report. (23rd January 2019)

• Record Type: Journal Article
• Authors: Freire, Daniela L.; Frantz, Rafael Z.; Roos‐Frantz, Fabricia
• Is Part Of: Software, practice & experience. Volume 49:Number 5(2019)
• Page Start: 921
• Page End: 941
• Publication Date: 2019-01-23
• Subjects: enterprise application integration -- integration framework -- integration patterns -- integration platform -- message‐based middleware -- multiple‐criteria decision‐making -- orchestration engine -- performance evaluation; Computer software -- Periodicals; Computer programming -- Periodicals; Computer programs -- Periodicals; 005.3
• DOI: 10.1002/spe.2679
• Languages: English
• ISSNs: 0038-0644

Abstract

Summary: Enterprises turn to their software applications to support their business processes. Over time, it is common for a company to end up with a wide range of applications, which are usually developed in‐house by its information technology department or purchased from third‐party specialized software companies. The result is a heterogeneous software ecosystem with applications developed in different technologies and frequently using different data models, which brings challenges when two or more applications have to collaborate to support a business process. Integration platforms are specialized software tools that help design, implement, run, and monitor integration solutions that orchestrate a set of applications. The run‐time system is the component of integration platforms responsible for running integration solutions, which makes its performance a critically important issue. In this paper, we report our experience in evaluating and comparing four well‐known open‐source integration platforms in the context of a research project where performance was a central requirement to choose an integration platform. The evaluation was conducted using a decision‐making methodology to build a ranking of candidate platforms by means of subjective and objective criteria. The subjective evaluation takes into account expert preferences and compares integration platforms using the analytic hierarchy process, which has been used in many applications related with decision‐making. The objective evaluation is build on top of properties distributed on three dimensions, namely, message processing, hotspot detection, and fairness execution, which compose the research methodology we used. The evaluated platforms were ranked to identify the one with the best performance.