Bachelor Thesis, 2025
Supervisors: Mehran Salmani, Boris Koldehofe
Description
This thesis investigates the use of the Publish–Subscribe pattern as the communication middleware in a microservice application that collects and analyzes data from IoT measurement devices. The goal is to achieve scalable performance, particularly in terms of throughput. The prototype is implemented using Apache Kafka as the Publish–Subscribe platform, with a focus on determining the appropriate number of topic partitions and an effective partition-assignment strategy to sustain the target throughput.
Additionally, the thesis leverages the decoupled nature of the Publish–Subscribe model to separate two application workflows that consume the same data stream: (1) persisting measurement data to a database, and (2) processing and rendering that data for end-user visualization with low latency.