This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(195.2, 95%, 28%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EVS%3C/text%3E%3C/svg%3E)
I have machine learning models which is deployed in Azure aks in 'x' Cluster , Now i want to move this models and its endpoint in another cluster which is 'y' within same workspace and subscription ,So how to do this without any changes to its rest endpoint as this endpoints are in production use already.
also if this is not possible then can i upgrade my x cluster with newer version without affecting my endpoints
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(22.400000000000002, 11%, 12%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ERM%3C/text%3E%3C/svg%3E)
@Vishal Suryavanshi Thanks for the question. Here is link to the blog for New managed online endpoints features in Azure ML: Autoscaling, Debugging, MLflow and more.
@Vishal Suryavanshi I see that when the CI is turned off, the models deployed with this method needs to be deployed again in the next start. According to my tests, it takes ~20 sec to deploy a model to a port. Since model deployment ports are not persistent after restart, you may consider using a CI-CD pipeline to run deployment script after CI restart. You may keep relevant Compute Instance (VM) on 7x24 so that the web service can be reached always, or if use case is batch scoring, you may optimize the cost by turning the CI off according to a schedule (private preview feature). Or you may automate CI deletion & re-creation with a script running it according to a schedule. I also see that the scoring requests need to be authenticated, so you will need to use some type of Service Principle authentication if you use this option.
Thanks for your reply , but will i get information there regarding my question because i cant find anything there
ok got it , but if i deploy it again then with new compute target my rest endpoint address will change or it will be same .?
@Vishal Suryavanshi Please Leverage the goodness of managed online endpoints without extra code, will investigate and update on your approach.