Observability 101

Before session 1

Set up your laptopAccept your GitLab and Datadog invitations, install the tools and run the setup check.

Useful linksDatadog, GitLab, EpitaCard, and a Datadog Learning course for each session.

How the course works

You inherit a marketplace that is already in production, with customers placing orders every minute and almost no instrumentation. Your group adds one layer of observability per session, then uses all of it to find out what broke when incidents hit your stack.

1. 1Setup
   blind stack
2. 2Logs
   Logs
3. 3Metrics
   LogsMetrics
4. 4Traces
   LogsMetricsTraces
5. 5Connect
   LogsMetricsTracesSLOs
6. 6Incident
   LogsMetricsTracesSLOs

Your stack gains one layer of observability per session, then you use all of it.

Groups, labs and grading are on the course format page.

Sessions

1. 1. Why observability?Observability versus monitoring, the three pillars and environment setup.
2. 2. LogsSoonStructured logging, log pipelines and searching logs in Datadog.
3. 3. MetricsSoonMetric types, RED and USE, dashboards and a first monitor.
4. 4. TracesSoonDistributed tracing, context propagation and trace-log correlation.
5. 5. Connecting the pillarsSoonCorrelation, SLOs, error budgets and alerting that people trust.
6. 6. Incident responseSoonThe incident lifecycle and the final graded exercise.

What you hand in

Each challenge and incident ends in a write-up: a Datadog Notebook in your team’s organization, with live queries as evidence and a link to the merge request that holds your fix. The final exercise is written the same way, as a post-mortem. The write-ups page shows the structure.