FaaS execution models for edge applications

In this paper, we address the problem of supporting stateful workflows following a Function-as-a-Service (FaaS) model in edge networks. particular focus on data transfer, which can be performance bottleneck due to limited speed communication links some scenarios and propose three different schemes: pure FaaS implementation, StateProp, i.e., propagation application state throughout entire chain functions, StateLocal, solution where is kept local workers that run functions retrieved only as needed. We then extend proposed schemes more general case applications modeled Directed Acyclic Graphs (DAGs), cover broad range practical applications, e.g., Internet Things (IoT) area. Our contribution validated via prototype implementation. Experiments emulated conditions show applying locality principle reduces significantly volume network traffic required improves end-to-end delay performance, especially with caching nodes low link speeds.