Observability: metrics, logs and traces, from zero to DataCat

11 steps · about 5 h 32 min · Intermediate

  1. What is OpenTelemetry? (doc, OpenTelemetry, 10 min)

    It defines observability and the three signals (traces, metrics, logs) that every monitoring tool, DataCat included, is built around.

  2. Getting Started with Observability (guide, Checkly Docs, 15 min)

    A short beginner guide that ties the three pillars to the OpenTelemetry collector and to backends like Prometheus and Grafana.

  3. OpenTelemetry Primer: Observability from First Principles (guide, Matthew Reider, 45 min)

    It goes deeper on metric types, spans and logs, so you know what a platform has to ingest and store.

  4. Distributed Tracing Explained (video, YouTube, 12 min)

    Traces are the hardest signal to picture, and this video covers spans, trace IDs, context propagation and sampling.

  5. Lab 1 - O11y Primer (tutorial, O11y Workshops, 30 min)

    It covers instrumentation, resources and the API/SDK split, with a hands-on Flask example.

  6. Google SRE book: Chapter 6, Monitoring Distributed Systems (doc, Google SRE, 40 min)

    Open Chapter 6 from this contents page to learn what to measure (the four golden signals) and how to write alerts that aren't noisy.

  7. Getting started (doc, Prometheus, 30 min)

    You run a real metrics system, scrape a /metrics endpoint and write your first queries and rules.

  8. Getting started with Prometheus (tutorial, Opensource.com, 40 min)

    It adds a second hands-on pass with node_exporter, time series and rate queries to reinforce metrics.

  9. Quick install (doc, ClickHouse Documentation, 20 min)

    DataCat can store telemetry in ClickHouse, so this installs it locally with no account needed.

  10. Starter guides (doc, ClickHouse Documentation, 30 min)

    It teaches the SQL you need to create tables and query large volumes of log and trace rows.

  11. replit/datacat: Self-hosted observability (repo, Replit on GitHub, 1 h)

    This last step puts it all together: browse a self-hosted stack that accepts OpenTelemetry data and stores it in PostgreSQL or ClickHouse.