Microservice-Oriented Architecture for Battery Energy Storage Analytics : A Comparative Study of Implementation Approaches and Performance Evaluation

Sammanfattning

With the accelerated deployment of Electric Vehicles (EVs) and Battery Energy Storage Systems (BESS), the demand for scalable, secure, and interoperable battery data analytics becomes even more urgent. Current monolithic architectures can be ill-suited for distributed environments where real-time telemetry, multi-stakeholder data sharing, and fulfilling the obligations of emerging data sovereignty laws are a priority. In this thesis, we explore the potential application of microservice-oriented architectures for battery analytics through two supporting studies. The first paper presents a light-weight PowerShell prototype that emulates certain aspects of the Eclipse Data Space Connector (EDC) for testing secure data sharing ideas. Results show reaction times of sub-100ms, and a very stable behavior, which proves that microservice ideas can be evaluated with a minimum of work. The second work is a comparative performance study of monolithic and microservice architectures, utilizing Docker for evaluation and integration with the Arrowhead Framework for service orchestration and authorization. Our stress tests reveal that the two architectures provide similar throughput (~46 req/s), but the microservice one has had better latency consistency (65% less maximum latency) and better fault isolation. The thesis contributes with: (i) a prototyping approach for evaluating early-stage microservice ideas in practice, (ii) an empirical study to compare the performance of monolithic and microservices on realistic workloads, (iii) experiences on the integration of the Arrowhead Framework in the industrial battery analytics scenario, and (iv) practical guidelines to implement scalable and secure data-sharing platforms. These results show that microservices, while certainly adding operational overhead, look like a promising route to a durable and future-ready battery data ecosystem. Not only do the results have academic value but they also bear industrial significance. They propose that microservice-based architectures can address the challenges of predictive maintenance, interoperability between heterogeneous energy actors, and resilience for grid-connected storage systems. In addition, the findings can guide real-world deployment in energy-intensive contexts, including factories that manufacture and manage batteries.