Datadog

What it is

Datadog is a publicly traded (NASDAQ: DDOG) SaaS observability and security platform founded in 2010 by Olivier Pomel and Alexis Lê-Quôc. Rather than being a single tool, it’s a broad suite of integrated products sold under one platform: Infrastructure Monitoring for metrics and host/container-level telemetry, APM (Application Performance Monitoring) for distributed tracing across services, Log Management for centralized log collection, search, and analysis, and Continuous Profiler for code-level CPU and memory profiling in production. All four pillars, plus adjacent products like Synthetic Monitoring, Real User Monitoring, and Cloud Security, share a single agent, a single web UI, and a single tagging model, which is Datadog’s core pitch: correlate a metric spike with the trace, log lines, and even the specific function consuming CPU, without switching tools or re-mapping identifiers between systems.

Data is collected via the Datadog Agent (installed on hosts, containers, or as a Kubernetes DaemonSet) or via native cloud integrations (AWS, GCP, Azure) and OpenTelemetry, then ingested into Datadog’s hosted backend — there is no self-hosted or open-source version of the core platform.

Why teams choose it

• Genuinely strong cross-signal correlation — because metrics, traces, logs, and profiles are collected and tagged consistently by the same agent, pivoting from a dashboard to the underlying trace and log lines tends to work more smoothly out of the box than in a hand-assembled open-source stack.
• Very broad integration catalog covering hundreds of technologies and cloud services with minimal setup, which is a major time-to-value advantage for smaller teams without dedicated observability engineers.
• Cost is the most common complaint at scale — Datadog bills separately across products (hosts for infrastructure/APM, ingested and indexed volume for logs, profiled hosts for profiling), and usage-based billing across many dimensions can produce bills that scale faster than expected and are genuinely hard to forecast without active management.
• Some lock-in risk — while Datadog supports ingesting OpenTelemetry data, its deepest features (certain APM views, some correlation features) work best with the Datadog Agent and its own instrumentation libraries, which is worth weighing against a more vendor-neutral OpenTelemetry-first strategy.

Pricing model

Datadog uses usage-based SaaS pricing with per-product billing rather than one flat platform fee: infrastructure monitoring and APM are typically billed per host per month, log management is billed on a combination of ingested volume and indexed (retained/searchable) volume, and profiling is billed per profiled host, often as an add-on to APM. Most products offer a free tier with limited scope and a free trial, but production use at any real scale moves into paid tiers, and multi-product adoption (which is where Datadog’s cross-correlation value proposition is strongest) compounds the bill across several usage dimensions simultaneously.

Alternatives

For a similarly unified commercial platform, New Relic and Dynatrace are the closest direct competitors, each bundling metrics, tracing, logs, and (in Dynatrace’s case) profiling-adjacent features under their own platforms. Honeycomb offers a more focused, trace-first alternative built around high-cardinality event analysis rather than a full four-pillar suite. Teams wanting to avoid vendor lock-in often assemble an equivalent open-source stack instead, combining Prometheus or Grafana Mimir for metrics, Grafana Tempo or Jaeger for tracing, Grafana Loki or Elasticsearch for logs, and Grafana Pyroscope for profiling, all visualized through Grafana.