Skip to content

Grafana Dashboard

SketchLog ships with a pre-configured Grafana Dashboard to help you visualize your high-performance stream metrics out-of-the-box.

Features

The dashboard includes visualisations for: 1. Ingestion Rate: Tracks sketchlog_total_events to monitor the throughput of your streaming data. 2. Latency Quantiles: Tracks sketchlog_latency using your chosen percentiles (default: 50th, 95th, 99th, and 99.9th percentiles) emitted from the embedded DDSketch. 3. Estimated Unique Elements: Visualizes sketchlog_unique_count generated by the embedded HyperLogLog structures. 4. Sketch Memory Footprint: Tracks sketchlog_memory_kb to showcase the bounded, predictable memory usage of the underlying probabilistic data structures.

Importing the Dashboard

You can find the dashboard template inside the repository at dashboards/sketchlog-overview.json.

  1. Prerequisite: Ensure you have configured the Prometheus Exporter to emit metrics.
  2. Open your Grafana instance.
  3. Navigate to Dashboards > Import.
  4. Click Upload JSON file and select the sketchlog-overview.json file from the repository, or copy/paste the JSON contents directly into the "Import via panel json" text box.
  5. Click Load.
  6. At the top of the newly imported dashboard, click the dashboard settings (gear icon) and navigate to Variables. Ensure the DS_PROMETHEUS variable points to your correct Prometheus instance.
  7. Save the dashboard.

[!NOTE] The template uses a fixed UID (sketchlog-overview). If you already have a dashboard with this UID, re-importing this file will overwrite it in place.

Supported Metrics

For the dashboard to populate, your Prometheus instance must scrape the following metrics from the PrometheusExporter:

  • sketchlog_latency (Gauge, labeled by quantile)
  • sketchlog_unique_count (Gauge)
  • sketchlog_total_events (Counter)
  • sketchlog_memory_kb (Gauge)

Note: The dashboard uses the job and instance labels automatically assigned by Prometheus during scraping to allow you to filter across different servers or applications.