A Distributed Tool for Monitoring and Benchmarking a National Federated Cloud

Monitoring the health and benchmarking the performance of Cloud resources deployed at large comes with a set of interesting challenges. It calls for extensible, customisable and open-access tools, featuring control over monitoring tasks and configurations, probabilistic representation of measured performance, and declarative description of setups. In this article, we present an open-source monitoring prototype that targets all the above objectives by adopting a declarative task-oriented approach. We demonstrate how the prototype achieved TRL7, through its full integration into the production monitoring stack of the federated national Cloud powered by Consortium GARR, viz., the Italian National Research and Education Network.