Deploying your model
Deploying a model can be done in many ways, depending on the use case and data availability. For example, deployment may look like packaging a model in a container and deploying it on an endpoint or model that runs daily in a production workflow to provide predictions in tables that can be consumed by applications. Databricks has product features to pave the way to production for all inference types.
Model Inference
We’ve walked through the methods and tools that help you set up your model in production, and finally, you have a model ready for inference! But one key question you should consider as part of this process is how your model should be used. Do you need the results once a day? Is the model powering an application that requires real-time results? Your model’s purpose will help you decide the type of deployment you need. You’ve seen the words “batch” and “streaming” a few times in this chapter already...
