Skip to content

Click in-app to access the full platform documentation for your version of DataRobot.

Deploy a DataRobot model in a Portable Prediction Server

DataRobot AutoML models can be deployed to a containerized DataRobot prediction environment called a Portable Prediction Server (PPS). To deploy an AutoML model to a PPS, you can build models with AutoML, deploy a chosen model to an external prediction environment, and then deploy the model package in a PPS with monitoring enabled. Once deployed, you can monitor this portable model alongside models deployed in DataRobot prediction environments.

To create and deploy an AutoML model in a PPS, follow the workflow outlined below:

graph TB
  A[Register a model]:::link --> B{Create an external prediction environment?}
  classDef link color: blue;
  click A "./dr-model-pps-env.html#register-a-model"
  B --> |No|C[Deploy the model to an external prediction environment]:::link
  click C "./dr-model-pps-env.html#deploy-the-model-package-to-an-external-prediction-environment"
  B --> |Yes|D[Add an external prediction environment]:::link
  click D "./dr-model-pps-env.html#add-an-external-prediction-environment"
  D --> C 
  C --> E[Deploy the model package to a PPS]:::link
  click E "./dr-model-pps-env.html#deploy-the-model-to-a-PPS"

Register a model

DataRobot AutoML automatically generates models and displays them on the Leaderboard. The model recommended for deployment appears at the top of the page. You can register this (or any other) model to the Model Registry directly from the Leaderboard.

Register a model

Deploy the model externally to a PPS

The Portable Prediction Server (PPS) is a solution for deploying a DataRobot model to an external prediction environment. You can download the PPS from the developer tools and use it to deploy a model package from the Model Registry. Once running, the PPS installation serves predictions via the DataRobot API.

Note

Depending on the MLOps configuration for your organization, you may be able to download the PPS model package from the Leaderboard for external deployment. However, without associating the model package with an external prediction environment, you won't be able to monitor the model's predictions.

Optional: Add an external prediction environment

To create an MLOps model deployment compatible with the PPS, you must add the model package to an external prediction environment. Create an external prediction environment if you don't already have one in DataRobot.

Add an external prediction environment

Deploy the model package to an external prediction environment

To create an MLOps deployment with an external prediction environment, deploy a model package to an external prediction environment.

Deploy a model to an external prediction environment

Deploy the model package to a PPS

The model's PPS model package (.mlpkg) file and the command-line snippet used to initiate the PPS with monitoring are provided for any model tagged as having an external prediction environment in the deployment inventory. You can download the model's PPS model package and use the provided docker commands to deploy the model with monitoring enabled.

Deploy a model to a PPS


Updated July 26, 2022
Back to top