SYSTEM Cited by 4 sources

Datadog¶

Datadog is a SaaS observability platform covering infrastructure monitoring, APM, logs, synthetics, RUM (real user monitoring), and security. Datadog is itself a Tier-3 blog on this wiki (see companies/datadog); this page covers Datadog as a system adopted by other companies surfaced in ingested sources.

As a front-end / RUM observability substrate¶

GitHub's Files-changed-tab rewrite (systems/github-pull-requests) surfaces Datadog as the dashboarding substrate for INP tracking: per-interaction INP measurements, PR diff-size segmentation (so p95+ PR regressions don't hide under median-healthy aggregates), and memory tagging. (Source: sources/2026-04-03-github-the-uphill-climb-of-making-diff-lines-performant)
Figma's OpenSearch tuning (sources/2026-04-21-figma-the-search-for-speed-in-figma) surfaces Datadog as the metric source for the 8 ms avg / 1 s p99 mismatch — the dashboard surfaced per-shard latency as if it were end-to-end, hiding the 150 ms coordinator-view latency. Lesson: the observability substrate will happily show you the wrong number if you don't know what you're measuring.

As a metric source for an LLM-based RCA pipeline¶

Expedia STAR (systems/expedia-star) — reads Datadog metrics via the Datadog API as the input to a multi-step LLM-driven root-cause analysis pipeline. Ingested signals: inbound + outbound traffic / errors, HTTP + gRPC + GraphQL latency, container CPU + memory saturation, plus Kubernetes (container restarts, probe failures) and JVM metrics (heap, GC). Expedia calls out Datadog's rate-limiting as a design constraint STAR handles with "common resiliency patterns ... asynchronous operations and batch processing". Canonical wiki instance of Datadog metrics as the upstream telemetry for an LLM-based RCA pipeline that sits above Datadog's own dashboard layer. (Source: sources/2026-04-28-expedia-expedias-service-telemetry-analyzer)

As its own engineering org¶

Datadog publishes extensively on its own stack — see systems/datadog-workload-protection (eBPF runtime security), systems/datadog-mcp-server (LLM-agent observability surface), systems/bits-ai-sre (hosted SRE agent), and sources/2025-07-17-datadog-go-124-memory-regression (the Go 1.24 mallocgc bisect).

Stub¶

This page only catalogs Datadog's adopter-mentioned role in ingested sources. Datadog's internal architecture (event-store / timeseries backend / log pipeline / Husky / Duet) is covered by dedicated pages where ingested — e.g. systems/husky.

Seen in¶

sources/2026-04-03-github-the-uphill-climb-of-making-diff-lines-performant — GitHub's INP dashboard with PR-size segmentation + memory tagging as the client-side observability substrate.
sources/2026-04-21-figma-the-search-for-speed-in-figma — Figma's 8 ms-vs-1 s metric-granularity mismatch surfaced on a Datadog dashboard.
sources/2025-07-17-datadog-go-124-memory-regression — Datadog's own Go 1.24 RSS regression investigation (workload running on Datadog's infra).
sources/2026-04-28-expedia-expedias-service-telemetry-analyzer — Expedia STAR consumes Datadog metrics as the raw input to its LLM-based RCA pipeline; first wiki instance of Datadog in the upstream-of-an-LLM role.

companies/datadog — company / engineering-blog entry.
concepts/observability — the umbrella discipline.
concepts/interaction-to-next-paint — key RUM metric Datadog surfaces.
concepts/automated-root-cause-analysis — the discipline STAR applies on top of Datadog metrics.
systems/expedia-star — canonical wiki LLM-RCA consumer of Datadog metrics.
systems/datadog-workload-protection, systems/datadog-mcp-server, systems/bits-ai-sre, systems/husky — Datadog's own products / internals when covered by ingested sources.