Skip to main content

Use M3DB as remote storage for Prometheus

M3DB is an excellent candidate for a highly scalable remote storage solution for your Prometheus monitoring system. Many organizations are already using Prometheus and come to M3DB when they have outgrown their existing storage setup. With the ability to scale as needed and store large quantities of time series data, serving as backend storage for Prometheus is one of the main use cases for M3DB.

For a quick example to try things out, see the Prometheus getting started guide, which uses Prometheus to monitor itself as a starting point.

Configure M3DB to store data in Prometheus​

  1. Configure Prometheus to write data to remote storage. Copy the Service URI value from the Prometheus (Write) tab. In your Prometheus configuration file (mine is called prometheus.yml), add the following section, replacing <PROM_WRITE_URL> with the URL you copied.

    remote_write:
      - url: "<PROM_WRITE_URL>"

    Prometheus is now configured to send data to your Aiven for M3 service.

  2. To maintain access to the data using your existing Prometheus service, configure Prometheus with the remote_read configuration to read data from the remote storage. Copy the Service URI value from the Prometheus (Read) tab. In your Prometheus configuration file, add the following section, replacing <PROM_READ_URL> with the URL you copied.

    remote_read:
      - url: "<PROM_READ_URL>"
        read_recent: true

    The read_recent parameter makes Prometheus read all data from the remote storage; this is useful so that you can test that the setup is working. Without this setting, Prometheus will return the most recent data from local storage if it still has it there.

  3. Run Prometheus, and check that it starts successfully. After giving it some time to scrape a few metrics, you should see data in the Prometheus web interface. Verify that there is data in M3, either by querying the database directly or by using Grafana to visualize the data stored there.