Microservice Oriented Pipeline Architectures

Microservices are an architectural style becoming more and more popular nowadays. It's getting over old paradigma not only because of the benefits it provides but also because it fits perfectly with the Cloud world, in particular with the capabilities of the network to provide on-demand availability of computer system resources. The thesis goal is to design a microservice architecture starting from a monolith. The use case is a parsing application for generics semi-structured data coming from a huge variety of devices. Such devices can be for example IoT devices of a smart city system rather than an industry 4.0 sensor system. The devices generate data that need to be parsed, classified, processed, aggregated and eventually displayed in a human-readable fashion. In the context of application pipelines, the present thesis work aims to implement a microservice architecture addressing distributed system properties like scalability, maintenance, high availability, and fallback mechanisms in order to build a more robust and more agile architecture from the modularity point of view. However, Microservice Architectures are clearly not a cure-all. In fact, while it allows many advantageous features, there are drawbacks that need to be kept in mind when deciding to implement a Microservice Architecture like the structural complexity that such architectural style adds to the application. There is not a solution written in stone, this thesis shows which are the main challenges addressed when building Microservices in the context of pipelines and tries to summarize the advantages and disadvantages of the architectural design.