Quickstart: Deploy the Azure AI Video Indexer extension on Azure Arc

In this quickstart, you deploy Azure AI Video Indexer enabled by Azure Arc to an Azure Arc-enabled Kubernetes cluster by using the Azure portal or the sample deployment script. The steps configure the extension for recorded media uploads and insights on your edge infrastructure. If you want to enable real-time analysis for live video, see Manage Azure AI Video Indexer extensions for real-time analysis instead.

Prerequisites for Azure AI Video Indexer enabled by Azure Arc

Before you begin, make sure you have the following subscription approval, resources, permissions, and network configuration:

• An Azure subscription with an approved subscription ID. Sign up by using the form at Application for gated services before you deploy the extension.
• Permissions for a user to create Azure resources in the subscription.
• An Azure AI Video Indexer account. See Create a Video Indexer account.
• Hardware and software that meet the minimum requirements.
• An Azure Arc-enabled Kubernetes cluster. See How to connect your cluster to Azure Arc.
• A Kubernetes storage class that supports ReadWriteMany (RWX). Because indexing is I/O intensive, use storage with sufficient IOPS for your workload.
• If you plan to use a language model, label a node or node pool with workload:summarization by following Label nodes in an Azure Kubernetes Service cluster. The labeled machine must have at least 32 CPUs for production workloads. Intel CPUs are strongly recommended over AMD CPUs.
• An externally facing endpoint, which can be either a DNS name or an IP address. Set the endpoint as a secure transfer protocol (https://) and use it as the extension API endpoint. The Video Indexer web portal uses the endpoint to communicate with the extension. Use an ingress control to manage the endpoint. If the endpoint isn't publicly accessible, you can perform actions on the extension from the web portal only from the local network.

Deploy Azure AI Video Indexer enabled by Azure Arc in the Azure portal

Use the Azure portal to install the Azure AI Video Indexer extension on your Azure Arc-enabled Kubernetes cluster. The following steps configure the extension for media uploads and recorded-media insights rather than live video analysis.

1. In the Azure portal, go to your Azure Arc-connected cluster.

2. From the menu, select Extensions > + Add > Azure AI Video Indexer Azure Arc extension.

3. Select Create. The Create an AI Video Indexer extension screen appears.

4. On the Basics tab, under Project details, provide the following information:

   Field	Description
   Subscription	Select the Azure subscription for the extension.
   Resource group	Select the resource group for the extension.

5. Under Instance details, provide the following information:

   Field	Description
   Region	Select the Azure region for the extension.
   Connected K8S cluster	Select the connected Kubernetes cluster where you want to deploy the extension.
   Extension name	Enter a name for the extension.
   Video Indexer account ID	Select the Azure AI Video Indexer account that the extension connects to.
   Ingress endpoint	Enter the IP address or DNS name to use as the API endpoint.
   Storage class name	Provide a storage class supported by your Kubernetes distribution. For example, if you're using AKS, you can use azurefile-cli. For more information, see Storage Classes in AKS. If you're using another Kubernetes distribution, see its documentation for supported storage classes or instructions for providing your own.

6. Under Additional settings, for Content type, clear Live video and select Media uploads. This quickstart configures the extension for recorded media uploads and doesn't enable real-time video analysis.

7. Select Next to open the Processing + AI tab.

8. On the Processing + AI tab, provide the following information:

   Field	Description
   Agentic capabilities	Leave disabled for this non-live deployment. Agentic capabilities provide advanced video investigation capabilities and require two dedicated H100 GPUs.
   Summarization capabilities	Enable to use advanced vision event summaries for uploaded media. This capability requires a dedicated H100 GPU.
   Processing	Media uploads require one GPU unit for summarization capabilities. Live video and agentic capabilities aren't included in this non-live configuration.
   Toleration Key for GPU	Enter the toleration key for GPU nodes, such as nvidia.com/gpu, when you configure GPU workloads.
   Node selector (optional)	If there are multiple GPU node types, specify the node name and value for the summarization capabilities workload.

9. Select Next to open the Tags tab. Add tags as needed, and then select Next to open Review + create.

10. Review the configuration, and then select Create.

Manually deploy the Azure AI Video Indexer Azure Arc extension

Use the AKS cluster setup guide to manually deploy the extension. The guide provides end-to-end instructions for creating an AKS cluster, connecting it to Azure Arc, configuring and deploying the extension, and verifying the installation. Before you begin, have the following values available:

• Azure AI Video Indexer account ID.
• Azure Arc-connected Kubernetes cluster name and resource group.
• Extension endpoint.
• Storage class name.
• Extension name and version, if applicable.

Follow the guide to prepare the cluster and deploy the extension.

Use the Azure AI Video Indexer extension on Azure Arc after deployment

To use the extension in the Video Indexer web portal, open the extension's Media files tab and select Upload to upload and index a recorded video. For more information, see Watch live video recordings.

To automate uploads and retrieve insights, use the Azure AI Video Indexer enabled by Azure Arc APIs.

Next steps

• Watch live video recordings.
• Review the Azure AI Video Indexer enabled by Azure Arc APIs.