Datadog
Infra / DevOps · Usage-based
Overview
A monitoring and observability platform. It collects metrics, logs, and traces from servers, applications, and cloud services, and shows them on dashboards with alerts when something goes wrong. It was founded in 2010 in New York by Olivier Pomel and Alexis Le-Quoc, and went public on the Nasdaq in 2019. It started with infrastructure monitoring and has grown into a wide suite where each area, application performance, logs, real user monitoring, security, and more, is a separate priced product. It is one of the most widely used observability tools, especially at mid sized and large companies running complex systems across several clouds.
What people use it for
Engineering and operations teams use Datadog to watch the health of their systems in one place instead of many separate tools. They track servers, containers, and Kubernetes clusters with metrics and dashboards. They trace slow requests through a chain of services with application performance monitoring to find which part is the bottleneck. They collect and search logs alongside those traces. They watch real user data from browsers and mobile apps, including page load times and JavaScript errors. They run synthetic tests that act like a user and alert when a key flow breaks. They set alerts that page an on call engineer. Security teams use it to spot misconfigurations and threats across cloud accounts. During an incident, teams use it to find the cause quickly.
Key capabilities
Datadog covers infrastructure monitoring, application performance monitoring, log management, database monitoring, network monitoring, real user monitoring, session replay, synthetic testing, cloud security, CI and test visibility, feature flags, incident management, and AI agent observability. It passed more than 1,000 integrations in 2025, covering cloud providers, databases, and common software, and it accepts data through the OpenTelemetry standard. A feature called Watchdog flags anomalies automatically, and teams can track service level objectives, run incident management, and keep a software catalog inside the platform. Newer areas include LLM observability and cloud cost management. Pricing is modular and mostly per host per month: infrastructure monitoring starts around 15 US dollars per host, application performance monitoring is a separate per host charge, and logs are billed per gigabyte taken in plus a charge per million events, with retention priced on top. Real user monitoring is billed per thousand sessions. Add ons like custom metrics, high cardinality tags, indexed spans, and longer retention each add cost. There is a free tier for up to 5 hosts with one day of metric history, and a 14 day trial of paid features.
Limitations
Cost is the near universal complaint, often from people who otherwise rate the product highly. Because each capability is billed separately and add ons stack on top, bills grow fast as a system scales, and Datadog bill shock is a common phrase among engineering teams. The most frequent trigger is auto scaling: infrastructure grows for a real reason and the monitoring bill jumps with it. Custom metrics, high cardinality tags, serverless tracing, and indexed logs are all known sources of surprise charges. The pricing has many interacting parts and is hard to model in advance. Large deployments regularly run into six and even seven figures a year. The breadth also takes time to learn.
Insight
Datadog does the core job well. When systems are spread across clouds and services, having metrics, traces, and logs in one place with good dashboards and alerting is worth a lot, and the integration list means most tools connect with little effort. The catch is the bill, and it is a serious one. Every capability is a separate charge, add ons pile up, and cost tracks infrastructure size, so scaling up scales the monitoring bill with it, often without warning. Teams routinely describe opening an invoice that jumped for the month. Anyone adopting Datadog should decide up front what to monitor, watch custom metrics and indexed logs closely, and treat cost control as an ongoing task with an owner. It suits mid sized and larger teams with complex systems and the budget to match. Smaller teams often get enough from a cheaper or open source tool.
Pricing
Usage-based
Website
Last checked 2026-08-30