Suspend and resume Cloudera AI Inference service

Suspend and resume Cloudera AI Inference service by using the CDP CLI to scale Kubernetes pods and optimize public cloud compute costs.

To minimize cloud consumption costs, the Cloudera AI Suspend feature allows users to scale down the Kubernetes pods running on Cloudera AI infrastructure and CPU/GPU nodes for a specific Cloudera AI Inference service. When the resume operation is performed on the suspended Cloudera AI Inference service, the suspended pods automatically scale back up. Cloud consumption costs are a pain point for many public cloud users.

During suspension, all autoscaling node groups for the service are reduced to zero instances, with the exception of the Platform Infra node group. This reduction saves compute instance costs for the duration of the suspension. However, Kubernetes pods running on Platform Infra nodes remain active and continue to run even when a Cloudera AI Inference service is suspended.

While a Cloudera AI Inference service is suspended, you cannot access the Cloudera AI Inference service URL, and all associated model endpoints are unavailable. When the Cloudera AI Inference service is resumed, model endpoints automatically restore operations at their original URLs.

Suspending a Cloudera AI Inference service
To suspend a Cloudera AI Inference service, use the CDP CLI. Run the suspend-ml-serving-app command by providing the Cloud Resource Name (CRN) of the service you want to suspend:
```
cdp ml suspend-ml-serving-app --app-crn [***crn***]
```
Resuming a Cloudera AI Inference service
To resume a suspended Cloudera AI Inference service, use the CDP CLI. Run the resume-ml-serving-app command by providing the CRN of the service you want to resume:
```
cdp ml resume-ml-serving-app --app-crn [***crn***]
```