Description of problem: Telemetry is suffering of a bad reputation regarding performances, as first versions, before RHOSP9 were relying on MongoDB and Ceilometer API. It was performant enough to store the metrics but not to retrieve and exploit them – so the usage was very limited. Solution Overview: Gnocchi (for metrics) and Aodh (for alarms) have been implemented to resolve all scalabilities issues. MongoDB has been replaced by Gnocchi. Nothing specific needs to be deployed but we need to demonstrate and communicate on the new status of Telemetry: it now really scales! We should provide a dimensioning guide for Telemetry at scale, indicating how and how far it scales and performs.
Testing has been done by Alex Krzos and showed great improvements. He was able to monitor up to 20k VMs on a 5 minutes interval with OSP 12. Results are presented here: https://www.youtube.com/watch?v=PC96PpT05G4
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHEA-2017:3462