| Summary: | Latency sensitive IoT (Internet of Things) applications at the edge are designed using a microservice-based architecture. This architecture is comprised of a set of microservices, each implementing a simple functionality with clearly-defined interfaces, and applications are constructed by selecting and interconnecting appropriate microservices. To understand the performance implications of using a microservice-based architecture for constructing IoT applications at the edge, this paper provides a detailed evaluation based on an actual prototpye implementation and performance measurement. In our setup, an edge server fulfills dual roles of being an administrative controller of the IoT infrastructure and satisfying application’s latency and privacy constraints. We demonstrate the utility of this architecture by isolated and independent implementation of different microservices, constructing an IoT application by interconnecting these microservices, and potential sharing of microservices between different IoT applications running simultaneously to enhance interoperability. Finally, we provide an extensive performance evaluation focusing on application latency as well as CPU and memory consumption.
|
|---|