Container Service for Kubernetes:Best practice for Unity remote rendering

Step 1: Use Unity Editor to develop a Unity Render Streaming application

The version of the Unity Editor used in this topic is 2021.3.2f1c1/Mac M1 silicon.

1. Choose Windows > Package Manager in Unity Editor, and then choose + > Add package from git URL in the upper-left corner.
2. Enter and search for com.unity.renderstreaming@3.1.0-exp.2, and then click Add.
3. In the message that appears, click Yes to restart the project.
   After the project is restarted, the Package Manager page shows that the Unity Render Streaming package is imported.
4. In the Samples section of the Package Manager page, click Import to import the samples provided by Unity.
   In this topic, a sample of 1.19 MB in size is imported. After the sample is imported, the Assets directory is updated.
5. Choose File > Build Settings in the upper-left corner of the page, set Platform to Windows, Mac, Linux, Target Platform to Linux, and Scenes to WebBrowserInput, and then click Build. Rename the build as mac-linux.
   The following figure shows the directory after the build is created.
6. Choose Edit > Render Streaming > Download web app in the upper-left corner of the page to download the Unity application.
   In addition to compiling the Unity application, you also need to create a web server to stream the rendered graphics to browsers. In this topic, a sample provided by Unity is used. Therefore, you can directly download the web server that corresponds to the sample.
   After the downloading completes, an application that supports Remote Render Streaming is developed. The application consists of a Unity-executable file and a web server application. For more information about how Unity Render Streaming is implemented, see About Unity Render Streaming.

Step 2: Containerize and deploy the application

1. Containerize the Unity application and web server.
   1. Containerize the Unity application. For more information, see Best practice for Linux graphics applications.
      Refer to Editor and choose a base image version. In this topic, the ubuntu-2021.3.2f1-mac-mono-1.0.1 base image is used and the following dependencies are installed: vulkan-util, libc++1, and libc++abi1.
   2. Containerize the web server.
      Web server containerization is similar to web application containerization. The following code block shows a sample Dockerfile:

      FROM ubuntu:20.04
      WORKDIR /run
      COPY webserver ./
      CMD webserver -w

2. Create and deploy a pod.
   1. Create a file named unity-demo.yaml based on the following content.
      After you obtain the container images of the Unity application and web server, deploy them in the ACK managed cluster. The following code block shows an example of the YAML file. You can refer to Best practice for Linux graphics applications and modify the file.

      apiVersion: v1
      kind: Pod
      metadata:
        name: unity-demo
        namespace: default
      spec:
        # The Linux web server sample exposes the IP address 127.0.0.1 when it uses WebSocket. Therefore, browsers establish WebSocket connections to the elastic IP address (EIP) of the host when they access the host. 
        hostNetwork: true
        nodeName: xxx
        containers:
        - image: xxx:xx
          name: unity
          command: ["/run/mac-linux/mac-linux.x86_64"]
          securityContext:
            privileged: true
        - image: xxx:xx
          name: webserver
          ports:
            - containerPort: 80
              protocol: TCP
        restartPolicy: Always

      Two containers are created in the pod to share the same network namespace. The Unity demo establishes a WebSocket connection to the IP address 127.0.0.1:80 of the web server and the web server exposes port 80 to browsers.
   2. Run the following command to deploy the pod:

      kubectl apply -f unity-demo.yaml

Step 3: Access the application and verify remote rendering

Enter the host address EIP:80 into the address bar of your browser and click VideoPlayer Sample. The following result is displayed.