KATWARN—A Microservice-Based Architecture for Distributed, Flexible and Robust Warning Systems

Abstract

As in many other domains we witness in disaster management a general shift from isolated monolithic implementations towards highly robust, distributed tasking and processing in a systems of systems approach. Warning systems can serve as as representative example for a class of applications where this need for change can be witnessed. Currently, warning systems are mainly specific developments with a generally similar but unique and often monolithic architecture. For fulfilling these upcoming new requirements reference architectures for such system classes are necessary in order to prevent the current reinvention of wheels with each new implementation. This paper presents a reference architecture developed on the basis of several years of research and tested in the KATWARN warning systems. The approach has been tested in a wide area of application contexts ranging from mass application of a population warning system with approximately 4 million users in Germany and Austria to local target group-specific warning systems for critical infrastructures and large chemical industry plants. Based on the foundations of event-based and asynchronous architecture paradigms the implementation is described in a microservice-based reference architecture. The architecture demonstrates how to meet high performance, robustness and scalability requirements in distributed process environments.