Architecture of a system for early detection and prediction of mechanical malfunction of machinery by sound analysis. Analysis and comparison of available options for expanding or improving the existing system.

Abstract

The thesis describes a model for the choice between two architectural styles: microservices and a modularized monolith in the context of Industry 4.0, IoT software solutions. As up till now, the approach is not systematized with several open questions that leave a lot to interpretation. First, it is defined what microservices architecture is and the style is compared with more established ones, i.e., monoliths, modularized monoliths, and SOA. Practically, microservices architecture is still a niche approach. In most cases companies usually apply a hybrid approach or a form of containerization of existing solutions that are divided into containers not necessarily following the microservices style. As changing the architectural style coincides with the move to the off-premise cloud, frequently it is impossible to compare exactly the performance of the cloud-based solution with the previous on-premise version, as there are too many variables to consider. Thesis proposes a simplified model on how to approach such decisions, as well as explains the DDD technique for decomposition. Consequently, chosen microservices patterns are described, as well as how to monitor microservices.