Feature Overview:

Resource Optimization for OpenShift (both ros-ocp-backend and Kruize) is currently using PostgreSQL as the data store for metrics and recommendations. PostgreSQL is not exactly the best choice for metrics, even less at our scale.

Goals:  

• Migrate the time series data from PostgreSQL to Thanos (RHOBS) for ros-ocp-backend
• Migrate the time series data from PostgreSQL to Thanos (RHOBS) for Kruize
• Kruize and ros-ocp-backend share the same data, to avoid duplication and re-submittal of data
• Increase performance

Requirements:

• Thanos (RHOBS seems the most likely candidate)
• No need to move from CSVs to TSDB blocks on the operator side for now but do it if it will help

Background and Strategic Fit:

Better performance, scalability, etc. Money savings by using a better database type for this kind of workload.

Docs:

Probably not, since this should be an internal change.

SMEs and Stakeholders: