Data Store Architectures: Balancing Functionality and Performance

This study evaluates three distinct Data Store Architectures in the context of industrial manufacturing workloads. We compare generally promised properties of Polyglot Persistence, Managed Polyglot and Multi-Model architectures regarding performance and operational complexity with implemented instances of these architectures. We use a real-world manufacturing workloads to evaluate the implementations. Our analysis reveals that promised properties suggested by the literature are not always met in practice. Contrary to initial presumptions, Managed Polyglot and Multi-Model architectures can outperform Polyglot Persistence in terms of execution time, despite Polyglot Persistence's design for specialised use cases.