Generate dashboard views using Gemini

Database Center includes a feature called generative views, which uses Gemini to let you create customized dashboards by using natural language prompts. Instead of being limited to predefined layouts, you can describe the database fleet information and metrics that you want to view. Gemini processes these prompts to generate a dashboard layout of widgets and data visualizations that matches your request.

Generative views let you query your database fleet's state, configuration, and performance metrics. For example, you can request views showing resources with specific characteristics, such as resources with high CPU usage in a specific region, or databases that don't adhere to certain best practices. You can save the generated dashboards, which lets you return to these custom views to see updated data. You can also modify your original prompts to iterate on and generate new dashboards.

Before you begin

Create a generative view

To create a generative view, do the following:

  1. In the Google Cloud console, go to the Database Center page.

    Database Center

  2. In the navigation pane, click Generative views.

  3. Click New.

  4. In the Ask Gemini to build a dashboard field, enter your prompt. For a list of sample prompts, see Sample prompts.

  5. To preview the generated dashboard layout, click .

  6. To save the view, do the following:

    1. Click Save custom view.

    2. In the Dashboard name field, review the dashboard name and make any changes you want.

    3. Click Save.

Delete a generative view

To delete a generative view, do the following:

  1. In the Google Cloud console, go to the Database Center page.

    Database Center

  2. In the navigation pane, click Generative views.

  3. Click the generative view that you want to delete.

  4. Click Delete.

  5. To confirm the deletion, click Confirm.

Sample prompts

To help you write your prompts, you can use the following sample prompts to get started:

  • "Show me resources with high CPU usage in us-west1."

  • "Show me all the databases (all engines) violating configuration best practices for Availability and Data Protection."

  • "Show me a heat map of all production Cloud SQL Postgres instances where CPU utilization has exceeded 80 % in the last 24 hours. Filter for instances with at least 16 vCPUs."

  • "Display a widget of all AlloyDB for PostgreSQL, Cloud SQL and Spanner instances across the organization that don't have high availability (HA) features enabled and are currently running on a version older than the latest minor release."

  • "What are my top 5 prod Cloud SQL instances across the fleet by CPU utilization? Only include the instances which are higher than 8 vCPUs and 8 GB memory."