Observability Plane · Pipeline & Infrastructure Monitoring

Datadog

Monitoring and observability SaaS for infrastructure, applications, logs and data pipelines.

Overview

Datadog is the commercial all-in-one of this category: one hosted platform covering infrastructure metrics, logs, traces, profiles, real user monitoring, synthetics, database telemetry, cloud cost and security signals.

It is software as a service only. There is no self-hosted Datadog at any price, which is the first thing to check if your data cannot leave your own network.

Architecture is an agent per host, or a Kubernetes daemon set, plus per-language tracing libraries, with cloud integrations pulling metrics through provider APIs.

What genuinely differentiates it is that packaging and billing are the product. Each capability is a separately metered item, mostly per host or per gigabyte, and unusually for an enterprise platform the real rates are published.

Understand the billing mechanism before budgeting. Hosts are counted hourly and billed on the high watermark of the lower 99% of readings, so the top 1% of spikes are discarded rather than setting your bill.

Logs are decoupled in a way worth exploiting: ingestion is billed separately from indexing, so you can ingest, process, live-tail and archive without paying to index everything.

Its OpenTelemetry stance is now three-way, with its own collector distribution embedded in the agent, the upstream collector with a Datadog exporter, or direct protocol ingestion with no agent at all.

Tracing is sampled by default rather than complete, targeting ten traces a second per agent plus error and rare-span sampling, so a plan for what you keep matters at scale.

Its integration count is the largest here, published as more than a thousand, which is the practical reason teams standardise on it despite the cost.

The 2026 direction is agents and AI, with credits as a metered unit, a chat and agent builder, feature flags, and code security priced per committer.

Key features and capabilities

The same headings are used for every pipeline & infrastructure monitoring entry, so two tools can be read side by side.

What it collects
  • Infrastructure metrics, logs, traces and continuous profiles, each a separately billed product
  • Digital experience covering real user monitoring, session replay, product analytics and synthetic tests
  • Security signals including cloud SIEM, cloud security management, code security and supply chain
  • Specialist telemetry for databases, data streams, network devices, cloud cost and LLM observability
  • Error tracking and incident management as distinct products
How data gets in
  • An agent per host, or a Kubernetes daemon set, plus per-language tracing libraries
  • Its own OpenTelemetry collector distribution embedded in the agent, the documented recommendation
  • The upstream OpenTelemetry collector with a Datadog exporter, supporting tail-based sampling
  • Direct protocol ingestion with no agent, for serverless and constrained environments
  • Cloud integrations pull metrics from AWS, Azure and Google Cloud through provider APIs
Storage and retention
  • Hosted only, with no self-managed storage tier
  • Log ingestion is billed separately from indexing, so you can ingest and archive without indexing
  • Indexing is sold by retention policy, with a cheaper flex tier for long historical and audit retention
  • The free tier retains infrastructure metrics for one day
  • Cardinality limits are not published
Querying and dashboards
  • Dashboards, notebooks and per-product explorers for logs, traces and user sessions
  • Correlation across metrics, traces and logs in one interface is the core pitch
  • Indexing exists specifically to enable exploration, alerting and dashboarding over selected events
  • No single documented query language surfaced on the pages read
Alerting and incident response
  • Monitors with anomaly detection, outlier detection and forecast alerts
  • On-call as a per-seat product, with unlimited schedules and escalations
  • Incident management per seat, with automated timelines, and a bundle covering both
  • Cloud SIEM bills per event evaluated by a correlation pattern
  • Service level objective specifics are not published on the pricing pages
Ecosystem and standards
  • More than a thousand built-in integrations, the largest count in this category
  • Full protocol support makes it a backend for OpenTelemetry rather than a competitor to it
  • Prometheus-style metrics can be scraped through the agent's integrations
  • Grafana Enterprise ships a Datadog data source, so it can also sit underneath Grafana
  • Observability pipelines route and transform telemetry before storage, and forwarding sends logs onward
Where it runs and what it costs
  • Software as a service only; self-hosting is not offered at any tier
  • The agent runs on hosts, containers, Kubernetes and serverless platforms
  • Multiple regional sites exist, plus a federal offering, though the region list is not published
  • Billed mostly per host a month, with per-gigabyte, per-seat, per-committer and per-session units alongside

Pricing

SubscriptionFree for 5 hosts; Pro at $15 a host

Published in full, which is rare here. Infrastructure is free for five hosts with one day of retention, then $15 a host a month on Pro and $23 on Enterprise. Application monitoring runs $31 to $40 a host. Logs cost $0.10 a gigabyte ingested plus $1.70 per million events indexed, with flex storage far cheaper to retain. On-call and incident management are $20 and $30 a seat, database monitoring $70 a host, and code security $25 a committer. A 14-day trial needs no card.

Vendor pricing page →

Demos and videos

About Datadog

Datadog was founded by Olivier Pomel, its chief executive, and Alexis Lê-Quôc, its chief technology officer, who worked together before starting it, and is listed publicly on Nasdaq as DDOG. It is proprietary commercial software with no self-hosted edition, though notably the Datadog Agent itself is Apache 2.0, so the client is open source while the backend is not. It publishes quarterly results and customer counts as a listed company, and reports more than a thousand built-in integrations. There is no user-visible platform version, since it is continuously delivered.

datadoghq.com

Other pipeline & infrastructure monitoring tools

Azure Monitor

Observability Plane · Pipeline & Infrastructure Monitoring

Azure service for collecting and analysing telemetry from cloud and on-premises environments.

  • Cloud service

Grafana

Observability Plane · Pipeline & Infrastructure Monitoring

Open-source dashboards for metrics, logs and traces, with a managed cloud offering.

  • Open core

OpenTelemetry

Observability Plane · Pipeline & Infrastructure Monitoring

Vendor-neutral standard and tooling for collecting traces, metrics and logs.

  • Open source

Prometheus

Observability Plane · Pipeline & Infrastructure Monitoring

Open-source monitoring system and time-series database, a CNCF project.

  • Open source

Drafted with AI assistance and checked against the vendor’s own documentation.