Operating Predictive Services

Once a deployment is created and running, it is important to be able to monitor and manage its behavior and make sure it is operating as expected. With GraphLab Create, we offer the tools needed to monitor and operate Predictive Service deployments. The following sections explore several aspects of operating a predictive service:

The chapter about Monitoring and Metrics shows how to gain insight into the health and performance of a running predictive service. It includes sections about:

• Displaying the service's status in GraphLab Canvas,
• Using AWS CloudWatch to set alarms on specific metrics, and
• Using APIs to retrieve monitoring status and metrics.

Administration explains how a running predictive service can be reconfigured and managed:

• Starting and terminating,
• Changing the number of nodes and replacing nodes,
• Changing the cache configuration and behavior, and
• Modifying cross-origin request behavior.