Managing the Lifecycle of a Real-Time Service Deployment

Adding Deployment Configurations

For a deployed real-time service, add deployment configurations to match service changes and upgrade the service.

1. Log in to the ModelArts console and choose Model Inference > Real-Time Inference.
2. Choose More > Add Deployment in the Operation column of the target service.
3. Configure deployment information by referring to the parameter description in Configuring Deployment Settings.
4. Click Confirm Deployment and add a deployment task as prompted.

Starting a Service Deployment

You can start services in the Stopped or Failed state. Services in the Deploying state cannot be started. Start the service deployment by following these steps. Once started, the deployment will be in the Running state.

Stopping a Service Deployment

Deployments in the Running or Abnormal states can be stopped. You can stop them using one of these methods:

1. Log in to the ModelArts console and choose Model Inference > Real-Time Inference.
2. Click the target service name. On the service details page that is displayed, switch to the Deploy tab.
3. Locate the target deployment card and click Stop. In the displayed dialog box, click OK to stop the deployment.

Switching the Real-Time Service Version

Change the real-time service version based on service changes.

1. Log in to the ModelArts console and choose Model Inference > Real-Time Inference.
2. Click the target service name. On the service details page that is displayed, switch to the Deploy tab.
3. Locate the target deployment card and choose ... > Change Version to switch the current version.
4. In the displayed dialog box, select the target version of the real-time service, click Switch to This Version in the Operation column, confirm the version comparison information, and click OK.

   You can also delete the deployed versions in the dialog box. The current version cannot be deleted.

Interrupting a Service Deployment

Interrupt a service deployment in the Deploying state to quickly stop it. You can interrupt a service deployment if it is severely faulty and needs to be recovered immediately, resources need to be quickly released to deploy a service with a higher priority, or fast iteration is required in the test environment.

• If a deployment in the Deploying state is interrupted, the deployment status changes to Stopped, related resources will be released, and the interruption operation will be recorded.
• If a deployment in the Upgrading state is interrupted, the deployment status changes to Running.
• If an exception occurs, such as unauthorized permission, invalid service ID, or version number does not exist, the API returns an error message.

To interrupt a service, follow these steps:

1. Log in to the ModelArts console and choose Model Inference > Real-Time Inference.
2. Click the target service name. On the service details page that is displayed, switch to the Deploy tab.
3. Locate the target deployment card and click Interrupt. In the dialog box that is displayed, click OK to stop the deployment.

Deleting a Service Deployment

If a deployment configuration is no longer used, you can delete it to release resources.

The deletion cannot be undone.

Deleting a Real-Time Service

If a real-time service is no longer in use, delete it to release resources.

• A service cannot be deleted without agency authorization.
• You are advised to delete the LTS logs and streams when you delete the service. This prevents additional fees incurred by the logs and streams.

The deletion cannot be undone.

Log in to the ModelArts console and choose Model Inference > Real-Time Inference. Delete the real-time service in any of the following ways:

• Choose More > Delete in the Operation column of the target real-time service. In the displayed dialog box, enter DELETE and click OK to delete the service.
• Select services in the real-time service list and click Delete in the upper left corner of the list. In the displayed dialog box, enter DELETE and click OK to batch delete services.
• Click the target service name. On the displayed service details page, click Delete in the upper right corner. In the displayed dialog box, enter DELETE and click OK to delete the service.

Other Deployment Operations

• Upgrading a Real-Time Service Deployment: To update a real-time service deployment, modify its configuration settings to include a new deployment version.
• Scaling a Real-Time Service Deployment: Once a real-time service is running, its resource needs might shift as services evolve. ModelArts allows you to deploy and scale these services dynamically. You can modify instance counts or resource settings to match current demand.
• Cloning a Real-Time Service Deployment: This feature allows you to duplicate an existing real-time service's deployment setup, adjust it as needed, and quickly create a new deployment.