Query Regional Core Usage by Resource

Question

I very frequently run in to the "Operation could not be completed as it results in exceeding approved Total Regional Cores quota" error. I understand what this means and how to increase my subscription's limit, but first I would like to investigate what services/resources are using all the vCPUs.

How can I query for the resources or services currently using vCPUs?

Documentation reviewed that was not helpful:
https://learn.microsoft.com/en-us/azure/virtual-machines/windows/quotas (Covers how to query for the overall usage and limit, I want a breakdown on what is contributing to usage)
https://learn.microsoft.com/en-us/azure/azure-portal/supportability/regional-quota-requests (I do not wish to increase my quota at this time)

Answer 1

This informatin should definitely be available in the azure billing details. Go to Cost Management --> Billing, you should be able to see all the resources that's used/provisioned and billed. This should give you a good idea of whats been used. You can also export the cost data to the lake and do the analysis there.

You can also get an high level view from Cost Analysis sectin in the portal which should be able to provide you details about which service, service type, etc. Based on the info you can further go into the resource groups and figure out why its been used. For instance a databricks instance max worker thread is set to a large number.

If you are using Databricks, you can turn on biling schema . This will create a new schema called system . This will start recording every single cluster and number of master and worker nodes created with date and timestamp. You can then later query the data to know which cluster was using high number of worker nodes. There are few tables which provide you the details, for example select * from system.billing.usage