Investigating the Performance of selected Data Storage Concepts for AutomationML Models

Abstract

Cyber-Physical Production Systems (CPPSs) engineering relies on the effective and efficient coordination and collaboration of participating engineers from various disciplines implying a proficient knowledge exchange between them. Au-tomationML (AML) provides a standardized, XML-based data format allowing to exchange engineering data, which receives more and more attention as an engineering data model. When using shared data exchange platforms, efficient data storage for AML models is success-critical in CPPS engineering. The purpose of this paper is to draft a novel, flexible evaluation framework in the context of AML model storage, modification, and retrieval and to evaluate two particular data storage paradigms, i.e., XML-(BaseX) and graph-based (Neo4J) databases. Based on a common best practice API, we developed a prototype solution enabling a flexible exchange of underlying data storage paradigms for AML models. We used an academic AML data set for the performance evaluation of two selected database engines for common storage tasks. First results showed that BaseX performs better for creating, updating, and deleting operations while Neo4J performs better for reading operations. While BaseX efficiently supports storing and retrieving AML models, we also observed querying limitations in the AML API. Nevertheless, in the context of AML data storage evaluations, the selected data sets, and the solution approach can be seen as an initial benchmark.