Deploying a model to an endpoint
To get predictions in Power BI, Power BI needs to send data to the model and get the result back to store it as a new column in the dataset. To accomplish this, the model needs to be deployed to a web service. When you train a model with AutoML, a web service is very easily created. You only need to specify the following:
- The name of the deployment
- The compute used to generate predictions: either Azure Container Instances (ACI) for small-scale deployments or Azure Kubernetes Service (AKS) for large-scale deployments
Both ACI and AKS are container orchestration services. ACI is Azure's proprietary service and is easier to use. AKS is based on the open source Kubernetes technology to orchestrate containerized applications. Even though Azure ML can manage and maintain the AKS clusters used for model deployment for you, it is better to use it when you have the expertise to set it up yourself as the management can be quite complex...
