Observability

Datadog

A commercial SaaS observability platform: metrics, logs, traces, APM and security in one place.

Official site

What it is

Datadog was founded in 2010 and has been a public company (NASDAQ: DDOG) since 2019. It is a fully managed service that unites infrastructure monitoring, APM, logs, RUM, synthetic tests and security.

The Datadog Agent is installed on servers and collects metrics and logs. There are hundreds of ready integrations with clouds, databases, queues and frameworks. Dashboards, alerts and correlation between metrics, logs and traces work out of the box.

The main downsides are the price, which grows quickly with data volume and host count, and vendor lock-in. Data can also be sent through OpenTelemetry to reduce dependence.

Use it when

  • You want a ready solution without deploying and maintaining your own stack.
  • Many services and clouds where unified correlation of metrics, logs and traces is valuable.
  • The team wants a fast start and hundreds of integrations.
  • There is a budget for a commercial tool.

Avoid it when

  • A tight budget: cost grows with log volume, custom metrics and hosts.
  • You need self-hosting or strict data location requirements.
  • You want to avoid vendor lock-in: Prometheus, Grafana and Loki offer an open alternative.

Key concepts

Agent
A process on a host or in Kubernetes that collects metrics, logs and traces and sends them to Datadog.
APM
Application Performance Monitoring: distributed traces, profiles and service dependencies.
Tags
Tags like env, service, version join data from different sources for filtering and alerts.
Monitors
Alert rules on metrics, logs, traces and synthetic checks.
Custom metrics
Your own metrics through DogStatsD; their count directly affects the bill.

Example

Running the Datadog Agent in Docker with APM and logs (the key comes from an environment variable).

Interview questions

  1. 1. What are the three pillars of observability?

    Metrics (numeric series), logs (events) and traces (request path). Together they answer what broke, where and why. Datadog links them by tags, so you can go from a metric to logs and traces.

  2. 2. Why can a Datadog bill grow sharply?

    You pay for hosts, indexed log volume, custom metrics with high cardinality and spans. Control it with log filters, sampling and limiting tags.

  3. 3. How do you reduce dependence on Datadog?

    Instrument apps through OpenTelemetry and send data through a Collector. Then the backend can be changed without rewriting code.

Comparisons with this technology

Related technologies

Logos belong to their owners and are used for identification only. Icons: Simple Icons (CC0), Devicon (MIT).

Explore the platform