Datadog Paid

Monitoring platform with an AI assistant that triages alerts across logs, traces and metrics.

Datadog is a monitoring platform that collects infrastructure metrics, traces, logs and RUM data in one product, then uses an AI assistant (Bits AI) to summarize incidents and triage alerts. Infrastructure monitoring starts at $15 per host per month; typical real bills land far higher.

What is Datadog?

Most teams start with one problem — “we can’t see why the checkout endpoint is slow” — and end up with five separate vendors. Datadog’s pitch is consolidation: the metric, the trace and the log line for the same request live in one screen with a shared tag.

Its agent runs on about 20 lines of install, covers Docker, Kubernetes, EC2 and bare metal, and reports host, container and process metrics without you writing scrape configs.

Product Features

  • Infrastructure monitoring: host and container metrics with 15-month retention on metrics.
  • APM and distributed tracing: service maps, latency percentiles, error tracking, span-level flame graphs.
  • Log management: ingest, parse, index and search, with Live Tail for real-time debugging.
  • LLM Observability: trace prompts, token usage and latency for AI features you ship.
  • Bits AI: asks follow-up questions during an incident and drafts a summary from telemetry and Slack threads.
  • 110+ integrations: AWS, GCP, Kubernetes, Postgres, Kafka, Airflow and most managed services.

Product Highlights

Correlation is the reason people stay. Clicking from a spike on a latency graph straight into the offending trace, then into the log lines from that span, saves an hour of tab juggling during an incident.

  • Anomaly detection on metrics, so alert thresholds adapt to weekly patterns.
  • Dashboards are shareable and embeddable outside engineering.
  • Synthetic monitoring checks your endpoints from 10+ global locations.
  • Service catalog plus on-call scheduling in the same tool.

Use Cases

  1. Kubernetes clusters where a pod restart loop needs fast root-cause tracing.
  2. E-commerce checkout latency budgets tracked as SLOs.
  3. Incident response where the on-call engineer gets a summary instead of 40 alerts.
  4. Monitoring LLM features for latency, cost per request and prompt failures.
  5. Database performance analysis on Postgres and MySQL replication lag.

How much does Datadog cost?

Infrastructure Pro is $15 per host per month (annual); APM is $31 per host; log ingestion runs $0.10 per GB plus indexing at $1.27 per million events. Custom metrics above 100 per host cost extra, which is how bills get out of hand — teams have published $50k+ annual invoices after a metrics-tagging mistake.

Who should avoid Datadog?

Small teams and side projects. Grafana Cloud’s free tier, or Prometheus plus Loki, covers a few services for near zero. Datadog’s cost model rewards consolidation at scale and punishes small teams that forget to set log sampling. There’s also lock-in: dashboards and alert definitions don’t migrate cleanly, so leaving is a project.

Quick Summary

  • Datadog unifies infrastructure metrics, APM traces, logs and RUM.
  • Infrastructure starts at $15 per host per month; APM at $31.
  • Bits AI summarizes incidents and suggests next debugging steps.
  • Best for teams running 20+ services that need one correlated view.
  • Log and custom-metric pricing surprises are a real risk.