Long term metrics storage and consolidation
Prometheus is not very good at long term metrics storage, especially when there may be multiple Prometheus servers deployed for a single project/group. (For example, a cluster just for production.)
We need a way to consolidate this data, and archive it off for more efficient long term storage.
Thanos may be a great way to solve both of these problems: https://github.com/improbable-eng/thanos