Skip to content
Tutorials Center
My learning

Topics

Lesson 1 of 37 ยท Getting Started

What Is Datadog? Observability Platform Overview

Learn what Datadog is, which problems it solves, and how metrics, logs, traces, monitors and dashboards fit together in one SaaS observability platform.

Software that runs in production breaks in ways you did not plan for. A server fills its disk, a deploy raises the error rate, a database query gets slow only for one customer. Datadog is a SaaS platform that collects data from your servers, containers, cloud services and applications, stores it in one place, and lets you search it, graph it and get alerted when something looks wrong.

This course teaches Datadog as a working engineer uses it. A word about this site first: Datadog is a hosted commercial product, so it cannot run inside a lesson. Menus and commands are described in words, YAML and shell snippets are shown read-only, and every TryIt is a small Python stand-in that models an idea (tag rules, metric aggregation, log parsing). The stand-ins teach the logic; they are not Datadog itself.

The three pillars, and then some

Observability is usually described with three kinds of telemetry:

  • Metrics are numbers sampled over time: CPU percent, requests per second, queue depth. They are small, cheap and great for graphs and alerts.
  • Logs are timestamped records of events: one line per request, one stack trace per crash. They are detailed, but large.
  • Traces follow one request across services, showing which step took how long. Datadog's tracing product is called APM (Application Performance Monitoring).

Datadog adds products around them: dashboards to see data, monitors to alert on it, Synthetic Monitoring to test your site from outside, Real User Monitoring (RUM) for browser and mobile sessions, security products that look for threats in the same data, and cost tools for cloud and Datadog spending. The selling point is that these share one set of tags, so you can jump from a graph spike to the logs and traces from the same service in the same minute.

How data gets in

Data flows from many sources to Datadog's intake over HTTPS:

flowchart LR
    H["Hosts and containers<br/>(Datadog Agent)"] --> I["Datadog intake"]
    A["Your apps<br/>(tracing libraries, DogStatsD)"] --> H
    C["Cloud integrations<br/>(AWS, Azure, GCP APIs)"] --> I
    L["Direct API calls<br/>(logs, metrics, events)"] --> I
    I --> P["Metrics, Logs, Traces,<br/>Dashboards, Monitors"]

The four common paths into Datadog. The Agent is the most important one.

The Datadog Agent is software you install on each host. It collects system metrics, runs integration checks for software such as nginx or PostgreSQL, tails log files, and receives traces and custom metrics from your own applications. Cloud integrations pull data from cloud provider APIs without installing anything. You can also send data straight to the HTTP API.

What you pay for

Datadog is billed per product, and each product counts something different. Infrastructure monitoring is priced per host. Logs are priced per gigabyte ingested plus the number of events indexed. APM is priced per APM host or per container. Custom metrics are priced by the number of unique time series. Prices change, so always read the current pricing page. What does not change is the habit this course keeps repeating: know what each product counts, because that tells you what makes the bill grow. Chapter 8 covers cost control in depth.

A quick map of a real workflow

Imagine an online shop. At 14:03 the checkout error rate climbs.

  1. A monitor on checkout error rate sends a message to the on-call engineer.
  2. The message links to a dashboard showing latency and errors by version.
  3. The engineer filters logs to service:checkout status:error and sees a database timeout.
  4. A trace shows the slow SQL call, and the version tag shows the problem started with the 14:00 deploy.

Every hop worked because the data carried the same service, env and version tags. Tagging is the thread that ties the product together, so you will see it in every chapter.

Python Stand-in: which pillar answers which question?

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Python Stand-in: a tiny correlation by shared tags

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The second example is the whole idea in ten lines: because both records carry service and env, you can match them. Datadog does this at scale.

Where Datadog fits

You can also run open source tools (Prometheus, Grafana, Loki, OpenTelemetry collectors). The trade-off is simple: with Datadog you do not run the storage and query systems yourself, and the products arrive pre-integrated; in return you pay a subscription and your data lives in a vendor service. Many teams mix both, sending OpenTelemetry data to Datadog. See the Grafana tutorial for the open source side.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. Which telemetry type follows one request across several services?

    Traces (Datadog APM) record the path and timing of a single request through services.
  2. 2. What is the Datadog Agent?

    The Agent runs on each host or node, runs checks, tails logs and receives traces.
  3. 3. Why do shared tags such as service and env matter?

    Shared tags are the key that correlates metrics, logs and traces.
  4. 4. Which product alerts you when a metric crosses a threshold?

    Monitors evaluate data and notify people when conditions are met.

Summary

  • Datadog is a SaaS observability platform: metrics, logs, traces, dashboards, monitors, synthetics, RUM, security and cost tools.
  • The Agent, cloud integrations and APIs are the ways data arrives.
  • Shared tags (service, env, version) connect the products.
  • Each product bills a different unit, so know what you pay for.
  • Datadog cannot run on this site; stand-ins model the logic.

Next: Accounts, Sites and API Keys