Datadog

Visit Website

Introduction

Datadog is a hosted observability platform. An agent on a host, or a library in an application, sends metrics, traces, and logs to Datadog’s cloud. You query them in dashboards, monitors, and a trace explorer. The same account can add real-user monitoring, synthetic API and browser checks, CI Visibility for pipelines and tests, and security products. Each of those is a separate SKU on the public price list.

The platform itself is closed source. Datadog publishes an agent and some client libraries, but this directory entry is the commercial service at datadoghq.com, not one of those libraries. There is no single “Datadog seat” that includes every product. Infrastructure, APM, and logs are billed on different units.

Features

1. Infrastructure metrics: The agent reports CPU, memory, disk, and process data per host. Container and serverless views are additional products.
2. APM: Language libraries emit traces. The service map and trace search show where time went. APM is billed per host or per traced serverless invocation, not as part of the infra host price.
3. Logs: Ingest is priced per GB. Indexed logs, the ones you can search later, are priced per million events and depend on retention.
4. Monitors and alerts: A monitor watches a metric, a log query, or a trace, and pages a person or a chat channel.
5. Dashboards: Graphs and query widgets over the same data. Useful, and easy to duplicate until someone owns the board.
6. Synthetics: API and browser checks from Datadog-managed locations. This is the piece that overlaps a test tool: a scripted browser journey on a schedule, billed as a synthetic test, not as a CI suite.
7. CI Visibility: Pipeline and test spans, including flake signals, if you install the CI integration and test hooks. It reads the tests your runner already executed.
8. Security and cloud cost SKUs: CSM, App and API Protection, and Cloud Cost Management sit in the same account and on the same invoice style. Turn them on deliberately.

Pros

1. One place for the three pillars: Metrics, traces, and logs can be pivoted from the same host or service name, which is the reason teams accept the SKU list.
2. The agent is widely documented: Kubernetes, ECS, and the major clouds have a known install path. You are rarely the first to wire a runtime.
3. CI Visibility attaches to existing tests: Playwright, JUnit, and pytest can emit spans without replacing the runner.
4. Synthetics cover the production URL: A check every few minutes finds a down checkout that CI, pointed at staging, will not.
5. Public list prices exist: Infrastructure Pro is on the price list, so a first estimate does not require a sales call. The estimate gets worse as you add SKUs, but the unit prices are published.

Cons

1. The invoice is a stack of meters: A host with infra, APM, and indexed logs is three charges. Custom metrics are another. Teams get surprised when a high-cardinality tag explodes the custom-metric bill.
2. Not a test framework: CI Visibility and synthetics observe or schedule checks. They do not replace Playwright, pytest, or JUnit.
3. Query language and retention are proprietary: Dashboards and monitors do not move cleanly to Grafana or another backend. Log retention you did not buy is data you cannot search.
4. Agent overhead: The agent and trace libraries use CPU and add latency if sampling is left at 100% on a busy service.
5. Closed platform: You cannot self-host Datadog. A network partition between your VPC and Datadog is an observability outage.

Usage and Links (Example)

The usual first install is the agent, not an application change. On a host, the docs give a one-line install that registers DD_API_KEY and starts the agent. In Kubernetes, the Helm chart datadog/datadog is the path Datadog documents. The API key is a secret. The site the agent talks to depends on your region (datadoghq.com, datadoghq.eu, and others).

For traces, add the language library and point it at the agent:

pip install ddtrace
ddtrace-run python app.py

That instruments common frameworks. You still configure service name, env, and sample rate. Do this in a non-production environment first and look at one trace before you roll it out.

Resources:

• Datadog website
• Documentation
• Pricing
• Price list

Pricing

Datadog publishes a price list rather than one plan. Figures below are the infrastructure line, checked against that list. Everything else is extra.

1. Infrastructure Pro: $15 per host per month billed annually, or $18 month-to-month and on-demand.
2. Infrastructure Enterprise: $23 per host per month billed annually, or $27 month-to-month.
3. APM, logs, synthetics, CI Visibility, RUM, and security are separate rows. Examples on the same list: logs ingestion at $0.10 per GB, container monitoring, and serverless APM per million traced invocations. Read the row that matches the product you will turn on.
4. Trials: Datadog offers a time-boxed trial. The ongoing free allowance, if any, is small and product-specific. Do not plan a production account around a trial.

Custom metrics, indexed logs, and high-cardinality tags move the bill more than the host count in many accounts. Ask for the price list row before enabling a product in the UI.

Recommended for?

1. Platform teams who want metrics, traces, and logs in one commercial backend and can staff the bill.
2. Kubernetes estates that already match Datadog’s Helm install and tagging scheme.
3. CI leaders who want pipeline and test traces next to production traces, via CI Visibility, without changing the test runner.
4. Not for a single service on a tight budget unless you have counted hosts times SKUs. A self-hosted Prometheus and Grafana stack is the smaller starting point.

Conclusion

Datadog is a broad commercial observability platform with public unit prices and a long list of optional products. Infrastructure Pro at $15 per host per month is the start of the quote, not the quote. Use synthetics and CI Visibility as signals around the tests you already have. Turn on APM and log indexing only for the services you intend to pay for.