How is Azure ML instance cost calculated?

Martin Korytak 6

Hello,

I've created an endpoint for scoring using a modified version of this tutorial: https://learn.microsoft.com/en-us/azure/machine-learning/how-to-deploy-existing-model. I used this command for specifying resources: AciWebservice.deploy_configuration(cpu_cores=1, memory_gb=12) and my resource group is "West EU". I'd like to know how the cost is calculated. I assume that the requested resources are converted to an instance (is that right?). I have found this useful website calculator but I cannot find which instance I am using.

How can I retrieve the information? Can I also do it programatically?

Many thanks.

1 answer

Ramr-msft 17,616 Reputation points

2020-11-30T13:35:11.803+00:00

@Martin Korytak Thanks for the question. On our website all the prices of Azure ML service are displayed and FAQ at the end.
https://azure.microsoft.com/en-us/pricing/details/machine-learning/
Yes, when you deploy your model to ACI with a default environment you can customize the deploy configuration (i.e. the number of cores and amount of memory made available for the deployment) using the AciWebservice.deploy_configuration().
Please sign in to rate this answer.
Ramr-msft 17,616 Reputation points

2020-11-30T13:38:54.027+00:00

Right now you can expand view to see a breakdown of cost by service in cost management. Also You only pay for Azure resources used, such as compute and storage. There is no additional charge to use Azure Machine Learning.

Martin Korytak 6 Reputation points

2020-11-30T14:14:59.103+00:00

@Ramr-msft thank you for your answers. Nevertheless, there is still one thing unclear. I've checked the website with pricing and at the end there is in FAQ:

Inferencing:
As a specific example, let’s say you deploy a model for inferencing all day for a 30-day billing month using 10 DS14 v2 VMs on an >Enterprise workspace in US West 2. For a billing month of 30 days, your bill will be as follows:

Azure VM Charge: (10 machines * CHF 1.177 per machine) * (24 hours * 30 days) = CHF 8,474.282

Azure Machine Learning Charge: (10 machines * 16 cores * CHF 0 per core) * (24 hours * 30 days) = CHF 0

Total: CHF 8,474.282 + CHF 0 = CHF 8,474.282

My question is: how do I know after the deployment which VM is used and what the cost is for a single machine in particular? In the provided example it is: DS14 v2 VM, CHF 1.177, respectively. This is the last tile which is missing in my puzzle to be able to estimate the cost.

Thank you.

Ramr-msft 17,616 Reputation points

2020-12-01T06:15:11.473+00:00

@Martin Korytak Thanks for the details. we are investigating the issue, Meanwhile please share your expand view to see a breakdown of cost by service in cost management for machine learning.

Martin Korytak 6 Reputation points

2020-12-01T09:44:25.7+00:00

@Ramr-msft here are the details of the service: container instances - 6.48 CHF, container registry - 0.42 CHF and the rest (storage, key vault, bandwidth) sums up to 0.01 CHF approximately. Please keep me updated on the status of this issue.

Thank you.
Sign in to comment