Cloud applications are the visible layer of cloud computing stacks. But to fully understand cloud applications and their performance, we need to look at the stack from the bottom up, starting from the hardware through to the management layer, the virtualized network and finally up to the application level.  So what is the best way to monitor an Openstack deployment?  What should be monitored? Which metrics are most relevant in our infrastructure and how can we efficiently locate and address bottlenecks?

This session is going to explain best practices of using telemetry from the perspective of the Cloud User, Administrator and Data Centre Owner, using real world examples which have been developed by the Mikelangelo EU project. It will explain the concept of telemetry and some of the limitations of existing OpenStack telemetry platforms. It will show how to deploy a telemetry framework, how to manage the thousands of metrics which are gathered, and how to derive insights from the collected data.  Real-time anomaly detection and Machine Learning algorithms are used to help us close the loop and provide output for the dynamic scaling and resource allocation algorithms which have become the norm in private and public Clouds.