Datadog

Overview

Datadog is a cloud-native observability platform that ingests metrics, traces, logs, and user experience data from your entire stack—applications, containers, servers, databases, third-party services—and surfaces them in a unified interface. Who uses it: Platform operators managing distributed systems, SRE teams triaging incidents across 50+ services, founders who can't afford to lose 30 minutes to a P1 because logs are scattered across five tools. The core workflow looks like this: An API endpoint starts timing out. Your alert fires in Datadog. You jump to the APM tab, see that a specific microservice is slow. You correlate that with infrastructure metrics (CPU spike on that host), then check the logs from that service during that exact minute. You find the culprit—a runaway database query—and roll back the deploy. MTTR: 8 minutes instead of 45. Datadog's distributed tracing follows requests across your entire microservice mesh. You see exactly where latency lives. The infrastructure monitoring catches resource exhaustion before it cascades. Log aggregation handles thousands of containers without choking. Synthetic monitoring runs API tests from multiple regions on a schedule, catching degradation before users do. Real User Monitoring (RUM) tracks actual page load times and interaction latency in production—not lab numbers. The pricing model is consumption-based: you pay for ingested metrics, logs, and trace spans. A typical mid-market SaaS operator with 20-30 services and moderate traffic runs $2k–$8k/month depending on data volume and retention. Startups often start smaller (a few hundred dollars), but scale hits fast once you're logging everything and running APM on all services. There's no published per-unit pricing—it changes, and Datadog's sales team negotiates volume discounts. Alternatives exist. Prometheus + Grafana is free but requires you to run and maintain the stack. New Relic offers similar APM but with different pricing and UI philosophy. Splunk is enterprise-grade but overkill for most SaaS teams. Honeycomb and Lightstep focus on observability-as-philosophy (cardinality, high-cardinality dimensions) rather than breadth. CloudWatch (AWS) or Azure Monitor are cheaper if you're all-in on one cloud, but they lack the cross-cloud and third-party integration depth. Honest limits: Datadog's UI is feature-rich but dense—new users often need 2–3 weeks to stop clicking randomly. The bill can surprise you if you're not careful about log sampling and trace retention policies; a misconfigured service can double your monthly spend overnight. Custom metrics are expensive. The onboarding assumes you know what you're monitoring; it's not a hand-holding tool. If your infrastructure is simple (one monolith, one database), you're paying for capability you don't need.