Summary:

In order to make informed decissions about changes in architecture in Watcher we need to have measures for metrics related to the decision-engine performance, specially:

• Size of the ClusterModel
• Time taken since the ongoing audit is created until execution starts
• Execution time of audits

Initial proposal is to use osprofile library as it's aligned with the rest of OpenStack services.

Goal:

• Define the high level implementation for the feature

TimeBox:

• Specify the maximum amount of time we want to invest 

Deliverables/Outcomes:

• Upstream blueprint describing the high level implementation.
• If needed local PoC of some relevant use case.