If you are seeing a D2a v4 Virtual Machine listed in the cost analysis, but can't find it in your Azure portal, it could be a part of a Machine Learning online endpoint/deployment or another managed service.
This article shows how to view costs for Azure Machine Learning managed online endpoints. Managed online endpoints are different from other resources since they must use tags to track costs.
Cost analysis view allows you to monitor costs at the endpoint and deployment level.
You can check the deployment configuration settings. Go to Azure Machine Learning > Endpoints and select the online endpoint you created. Look at the deployment settings, which may reference a compute target like a specific VM or managed cluster.
You can also check Under Azure Machine Learning Workspace
- Machine Learning Compute: When you deploy a model as an online endpoint, it may use compute resources that are not always visible in the Virtual Machines section.
- Go to your Azure Machine Learning Workspace in the portal.
- Under Compute, check the following sections:
- Compute Clusters: This is where you might have scaling compute resources set up for your models.
- Compute Instances: These are individual machines for training and inference.
- Inference Clusters: Check if this cluster is using a D2a v4 VM behind the scenes.
The online endpoints could be using a compute instance or cluster that internally runs on a D2a v4 VM, but it's managed and abstracted from the VM list.
Hope this helps!
Let me know if you have any further queries!
If the answer is helpful, please consider accepting answer and click "upvote".