Foundations
What Datadog is, the Agent and integrations, unified service tagging, and a tour of the UI. Start →
Datadog — From Zero to Hero
A standalone, developer-focused course on Datadog — the unified observability platform. From the Agent and unified service tagging to metrics, log management, APM, and the correlation model that ties them into one view per service — taught through real Agent config, DogStatsD, and Terraform. Every layered detail (ingestion vs. retention sampling, exclusion filters vs. archives) is verified against current docs.
What you’ll learn
Section titled “What you’ll learn”Infrastructure & Metrics
Agent-collected metrics, custom metrics via DogStatsD, Metrics without Limits, and dashboards. Measure →
Log Management
Collection and processing pipelines, indexes vs. exclusion filters, log-based metrics, and archives. Collect →
APM & Distributed Tracing
Traces, spans and the service map, ingestion vs. retention sampling, trace metrics, and profiling. Trace →
Correlating Logs, Traces & Metrics
Trace-log injection, unified service tagging in practice, the Service Catalog, and RUM context. Connect →
Monitors, Alerting & SLOs
Monitor types and thresholds, composite monitors, SLO error budgets and burn rates, and Synthetics. Alert →
Production & Ecosystem
Datadog as code with Terraform, cost governance, security basics, and a production rollout checklist. Ship it →
How Datadog Works
Inside the Agent (core, Trace, Process, Security and Cluster Agents, check scheduler, forwarder), DogStatsD aggregation and DDSketch, the trace sampling pipeline layer by layer, log pipeline processor semantics, and OTLP ingest — proven against a real Agent 7.83.3 and a local fake intake. Look inside →
Reading AI-generated Datadog
A 27-bug catalog of what AI-written Agent config, tracer setup, monitors and Terraform get wrong, a review checklist, the verification-tool ladder (agent configcheck, tracer config dump, terraform validate, API client models, the course harness), and three worked reviews — every bug proven. Review →
How this course works
Section titled “How this course works”Layered pipelines, not one dial
Metrics, logs, and traces are separate pipelines, each with its own ingestion, retention, and cost controls — this course teaches every layer instead of treating “sampling” or “retention” as one setting.
Current behavior, verified
Every claim is proven with the real tools — agent configcheck/status on Agent 7.83.3, tracer config dumps, a local fake trace intake, terraform validate and the API client models — or quoted from the official docs with a URL — including the split between log ingestion and indexing, and APM’s ingestion-vs-retention-filter distinction that a lot of tutorials collapse into one.
One tag triple, three pipelines
Not just “add some tags” but the full picture: how env/service/version — unified service tagging — let you pivot between a service’s metrics, traces, and logs instead of hunting through three disconnected tools.
Diagrams that explain
The Agent’s data paths, the log pipeline from intake to index, APM’s ingestion/retention split, and the SLO error-budget model as theme-aware Mermaid diagrams. Bilingual (English / ไทย) with saved quiz progress.
Verified against (2026-09-30)
Section titled “Verified against (2026-09-30)”Agent 7.83.3 (datadog/agent:7) · dd-trace (Node.js) 6.17.0 · ddtrace (Python) 4.15.2 · @datadog/browser-rum 7.15.0 · Terraform datadog provider 4.22.0 (datadog_monitor, datadog_dashboard, datadog_service_level_objective) · datadog-api-client (Python) 2.61.0 · Datadog Operator v1.30.0 / Helm chart datadog 3.250.3 · datadog-agent:7 (Docker/Helm) · Kubernetes Autodiscovery annotations in the current ad.datadoghq.com/<container>.checks v2 format · DD_TRACE_SAMPLING_RULES (works in both tracers; DD_TRACE_SAMPLE_RATE is not deprecated but is ignored by ddtrace 4.15.2).