How to get the cost breakdown per workload for different nodes

[altruistcoder]

Hello,

My team is exploring Opencost and Kubecost for a use case and we wanted to know whether Opencost has support for calculating costs for the clusters running on the OVHCloud managed Kubernetes service?

Also, while going through the Opencost specification, I found that Opencost calculates the overall cost based on the individual ratio of the CPU, GPU, and RAM of the underlying infrastructure (in case it is not already provided by the cloud provider) as given under the Appendix A here:

https://github.com/opencost/opencost/blob/develop/spec/opencost-specv01.md#appendix-a

But, I didn't quite understand, how we can get or calculate the ratio of these resources for the instances provided by any other cloud providers like OVH Cloud.

So, can anyone please share some documentation or references where we can get more info on how to calculate the per workload costs for the instances provided by any cloud provider?

[mattray]

I discussed adding support with the OVHCloud team at KubeCon EU earlier this year and they expressed interest, but they haven't followed up. I'm not familiar with OVHCloud's billing, but as the OpenCost Specification points out, we find the cost of each node in the cluster and split the usage of each container running on it to determine costs based on their requests and usage characteristics.

Here is a doc that covers the basics: https://docs.google.com/document/d/1JVPCeQmvGU88aw0bJJjtTzdl5RhhOqT7FPs2szuHBkc/edit

Feel free to reach out to me via email if you are planning on working on this, I'll see if I can get you some assistance ([email protected]).