By Nuha Alshuqayran
MSAs are often seen as the holy grail for many problems, like scalability and performance. However, it also introduces many caveats. For example, it introduces a myriad of dependencies.
In the problem at hand, they assume that the microservice implementation (deployment) and architecture model differ, and that mapping rules exist between the two models. They used a first case study to discover several rules, which resulted in a repository of rules. In the next step, they used static reconstruction to find the class diagrams, and used the Zipkin tool to trace communication between the MSs. Next, they determined the architectural concepts and identified the technologies used to implement the different concerns relevant to MSA (according to literature). Finally, they clustered and integrated the results, resulting in a metamodel, and created mapping rules to populate the metamodel. In the next phase, they executed their rules on several other cases, to validate and refine their results. They claim to recover architecture models, but their paper did not show any actual architectures recovered…