D3.2 - First release and evaluation of the monitoring system

This document, developed by the AI-SPRINT project, describes the first release and evaluation of the monitoring subsystem. The final release and evaluation of the monitoring subsystem is due at M24.
The document describes the components involved in monitoring subsystems, as well as the results of the preliminary evaluation. The technological choices are taken considering the requirements elicited in AI-SPRINT Deliverable D1.2 Requirements Analysis. The document first introduces the functional and non-functional requirements of the monitoring subsystem. Next the focus is on an analysis and evaluation of the available open source tools. Finally, the document describes the innovative architecture that will be
developed within the AI-SPRINT project which allows monitoring data collection also in case of edge/cloud disconnection and logs collection. The Proof of Concepts (PoCs) developed so far, the tests conducted and a roadmap for the development activities are also described.
It is worth mentioning that the architecture proposed in this document passed preliminary evaluation. The PoCs we implemented cover the most crucial parts of the implementation and tests in laboratory environments have been conducted. The results indicate that the monitoring subsystem will pass the requirements towards scalability and performance with no issues. The most important goals for the next twelve months are the implementation of means to easily define custom metrics, the support for hierarchical
deployment and the automation of monitoring subsystem deployments and configuration.