Customizing Grafana dashboards to monitor engine performance

Grafana dashboards provide a powerful and flexible way to visualize and monitor the performance and health of your Presto (Java), Presto (C++), and Spark engines. This guide outlines how to import default dashboards, apply filters, and customize them to suit your observability needs.

Before you begin

Before importing dashboards, make sure:
  • A Prometheus data source is configured in Grafana. Refer to Grafana data sources documentation.
  • Required metrics are available from the data source.

Procedure

Follow the steps to create default dashboards in Grafana UI for OpenTelemetry data:

  1. Importing the default JSON template

    You can import a predefined JSON dashboard template downloaded from Milvus, Presto (C++), Presto (Java), and Spark using the following method:

    1. After downloading the required JSON dashboard templates:
      1. In the Grafana UI, navigate to Home - Dashboards - New - Import Dashboard.
      2. Upload the dashboard JSON file.
      3. Select the appropriate Prometheus data source.
      4. After import, configure the filter instanceId by selecting the instance_id from the drop down and EngineId by selecting the engine ID (labeled as podUid) for your engine.
  2. Adding filters or Grafana variables

    To add a new variable:

    1. Open the dashboard.
    2. Click the gear icon and select Dashboard Settings.
    3. Navigate to Variables and select Add variable.
    4. Choose a variable type from the list (Query, Custom, Constant).
    5. Update the required fields and save.
    Refer to Grafana Variables for information.
  3. Customizing dashboards
    You can modify dashboard panels to better visualize your data:
    • Add new panels with custom queries and visualizations.
    • Edit existing panels to change metrics, queries, or visual settings.
    • Rearrange panels for clarity.
    • Update legends and labels.
    Refer to Grafana Build dashboards for information.