CODEWITHSUNDEEP
azure-aksazure-machine-learning-service
JakeUT
JakeUT

Reputation: 525

Azure Machine Learning Model deployment as AKS web service from multiple workspaces

I am trying to determine how the Az ML model deployment works with AKS. If you have a single AKS cluster but attach from two separate workspaces, will models from both workspaces get deployed into different azureml-fe's with different IP addresses OR a single azureml-fe with a single IP address? Reason I ask is because I want to purchase a certificate but am unsure if all the models (regardless of workspace) will get exposed under the same IP Address OR multiple IP Addresses? If its the former, I can do it with one certificate...otherwise I have to do it with multiple certificates or SAN based certificates. So if anyone has experience with this, please let me know!

Upvotes: 0

Views: 397

Answers (1)

Sairam Tadepalli
Sairam Tadepalli

Reputation: 1683

By checking AKS webservice class, we can do the multiple services links to single AKS cluster. The endpoint management was described in document, but this is representing one-to-one cluster and service. For multiple workspaces refer document.

Regarding azureml-fe. There will be one azureml-fe for one cluster. That means, when we are using different workspaces for deployment into one AKS, then only one azureml-fe and can be considered to take one certificate.

https://learn.microsoft.com/en-us/azure/machine-learning/v1/how-to-deploy-azure-kubernetes-service?tabs=python#azure-ml-router

Upvotes: 1

Related Questions