How to Use Datadog APM Traces to Find a Slow Endpoint
Enable APM, read a trace flame graph, spot the real culprit behind a p95 latency spike, and set an SLO that reflects what users actually…
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.
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.
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.
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.
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.