Persistent Volume Hourly Cost Metric

[Pokom]

So far we've focused our efforts on exporting metrics for the following resources:

• compute
• blob storage

One metric distinctly missing is tracking the hourly cost for persistent volumes. We would need two metrics:

• Hourly Cost of Storage for the persistent volume (rate)
• Amount(in GiB) allocated by the Persistent Volume Claim (usage)

These two metrics can then be joined together to calculate the total in spend (spend = usage * rate) for persistent volumes.

For both metrics, we need need the following labels:

• persistentvolume - The name of the persistent volume
• namespace - The namespace the pvc exists in
• cluster - The cluster the pvc exists in
• storage_class - Whether or not it's an SSD or HDD
• location - The region the pvc lives in

### Tasks
- [x] Implement persistent volume in GCP
- [ ] https://github.com/grafana/cloudcost-exporter/issues/235
- [ ] https://github.com/grafana/cloudcost-exporter/issues/236

[the-it]

Could we also just have one metric for each PVC, just cost per hour. If we need to scrape all PVCs, we also know there size, then we also know there total cost. Or is it somehow possible that a PVC is shared between namespaces?

[the-it]

• next question is this part of the GKE module?

[Pokom]

Could we also just have one metric for each PVC, just cost per hour. If we need to scrape all PVCs, we also know there size, then we also know there total cost. Or is it somehow possible that a PVC is shared between namespaces?

next question is this part of the GKE module?

I'll have a better answer to both questions once I start the implementation. It might end up being a dedicated module, but persistent volumes in particular are directly related to k8s resources so I think gke is where I'm going to start.

I like the notion of a single metric which takes into account exactly how much that pvc costs per hour as it simplifies things down stream in terms of queries. As long as there's no technical blocker, that's how I'll roll this out.

[Pokom]

Closing in favor of using existing issues. AWS is being implemented by @paulajulve(https://github.com/grafana/deployment_tools/issues/149203), and Azure still has to be planned out and picked up.