Learning Timeline
Key Insights
Key Metrics for Agent Monitoring
After deployment, focus on the following metrics in the Vertex AI dashboard to ensure your agent is performing well at scale:
- **Queries:** The number of requests received by the agent.
- **Latency:** The time taken by the agent to respond.
- **CPU & Memory:** Resource usage to assess if the resource allocation is sufficient.
- **Sessions:** Agent Engine manages user sessions automatically, and you can monitor session information here.
Step by Step
How to Deploy an Agent Using a Python Script
- In the deployment script file, ensure you have set the 'base requirements' needed for your agent to operate.
- Open your terminal or command line application.
- Use the 'cd' command to navigate to the directory where your project repository is stored.
- Run the deployment script by typing the provided command.
- Wait for the deployment process to finish. This script will automatically register your agent with the Vertex AI Agent Engine.
Monitoring Agent Performance on the Vertex AI Dashboard
- Log in to the Google Cloud Console and navigate to the Vertex AI page.
- In the left navigation menu, find and click on the 'Agent Engine' section.
- You will see a list of deployed agents. Click on the name of the agent you want to monitor.
- The performance dashboard for that agent will be displayed, showing key metrics.
