Monitoring In Function-As-A-Service Platforms

Functions deployed in FaaS platforms need to be monitored regarding resource consumption, errors, and application-specific metrics. Platforms like Google Cloud Functions or Azure Functions have dashboards and Web APIs that expose information about the execution of functions. However, the current approach collects data of general metrics about the function as an all and imposes a vendor-specific way for monitoring events. To the best of our knowledge, fine-grained function metrics are not available in FaaS platforms, e.g., the time that takes to execute only part of the function’s code. This work aims to explore how to build a system to provide fine-grained monitoring to FaaS platforms for developers. In this paper we present an architecture for this system, introduce the metrics library we are developing for FaaS platforms, and discuss some challenges.