Skip to content

SYSTEM Cited by 5 sources

Cloudflare Radar

Cloudflare Radar (radar.cloudflare.com) is Cloudflare's public Internet observatory: an always-on telemetry platform that publishes per-AS, per-country, per-prefix Internet data derived from Cloudflare's global edge visibility — including traffic patterns, DNS query shapes, HTTP adoption, IPv6 posture, attack trends, and BGP routing anomalies.

What it exposes

Radar surfaces several classes of data for any ASN or prefix:

  • AS profile — connectivity (upstreams / peers / downstreams), country, organization, traffic share at the Cloudflare edge.
  • Routing anomaly pages — per-event pages for detected BGP events (e.g. route leaks, hijacks) naming the leaker AS, the origin AS, the prefixes involved, and the neighbor ASes they were leaked to / from.
  • Traffic, adoption, and attack trends — aggregate metrics at country / AS / network granularity.
  • Historical alerting feed — cumulative route-leak alerts per AS, enabling baseline frequency analysis (e.g. "AS8048 has had eleven Type-1 hairpin route-leak events since the beginning of December 2025").
  • Internet Quality — (radar.cloudflare.com/quality) per-AS, per-country rankings of CDN providers on connection time, aggregated via trimean over RUM samples collected from Cloudflare-branded error pages against Cloudflare, Amazon CloudFront, Google, Fastly, and Akamai. Substrate for Cloudflare's periodic network-performance update posts. See patterns/comparative-rum-benchmarking.
  • Adoption of AI agent standards (AI Insights) — (radar.cloudflare.com/ai-insights#adoption-of-ai-agent-standards) weekly-refreshed chart scanning 200,000 most-visited domains for presence of each of the standards isitagentready.com checks — robots.txt, Content Signals, markdown content negotiation, MCP Server Cards, API Catalog (RFC 9727), Web Bot Auth directory, Agent Skills index, OAuth Protected Resource Metadata. Substrate for score-driven standard-adoption telemetry — turns the Agent Readiness Score from a diagnostic tool into a public adoption dashboard.
  • Response Status Code Analysis (AI Insights) — (radar.cloudflare.com/ai-insights#response-status) launched 2026-04-17. Distribution of HTTP status codes across AI-crawler traffic at Cloudflare scale — filterable by individual status code or by response class (2xx / 3xx / 4xx / 5xx). Available aggregated across all AI crawlers and broken down per-bot on the bot directory pages. Representative numbers from the 2026-04-17 launch: aggregate ~74 % 2xx, ~13.7 % 4xx, ~11.3 % 3xx, ~1.2 % 5xx; GPTBot-specific ~83 % 2xx, ~10 % 4xx, ~5.1 % 3xx, ~2.2 % 5xx. Filterable by industry (Data Explorer) and by crawl purpose (Training / Search / Assistant). Exposed via the Cloudflare Radar API. Framing (Cloudflare 2026-04-17): "status codes are ultimately how the web communicates policy to crawlers" — the surface turns the policy-lane framing of patterns/response-status-as-content-policy into a public web-wide measurement. Wiki's canonical instance of Radar as crawler-response-policy measurement substrate, sibling to the standard-adoption and BGP-forensics roles.

Route-leak alerting pipeline

The wiki's canonical instance of Radar as a substrate is the BGP-forensics use case from the 2026-01-08 Venezuela post (see sources/2026-01-08-cloudflare-a-closer-look-at-a-bgp-anomaly-in-venezuela). The pipeline auto-classifies BGP events against RFC 7908 route-leak types, assigns them stable IDs (e.g. leak-462460), and exposes a per-AS history page. Operators and researchers can then evaluate an alleged BGP anomaly against the AS's own baseline rather than against a global uniform prior — the disambiguator between "targeted BGP attack" and "chronic routing hygiene issue."

Why it works

The architectural preconditions that let Radar act as an Internet observatory — rather than just a Cloudflare-product dashboard — are ambient to Cloudflare's substrate:

  • Cloudflare's edge spans hundreds of POPs and sees a large fraction of global DNS / HTTP flows, giving per-AS traffic visibility without instrumented sampling partners.
  • Cloudflare peers with a large fraction of the world's networks (tens of thousands of peering sessions), so its BGP view is close to the global routing table.
  • Cloudflare runs anycast DNS / CDN, so per-client AS attribution is high-quality.
  • Combining DNS-level, HTTP-level, and BGP-level views in one observatory lets events (e.g. a BGP withdrawal) be joined to their user-facing consequences (e.g. DNS query rate drop).

Seen in

  • sources/2026-01-08-cloudflare-a-closer-look-at-a-bgp-anomaly-in-venezuela — canonical instance of Radar as BGP-forensics substrate: the per-AS route-leak history page for AS8048 (radar.cloudflare.com/routing/as8048) shows eleven route-leak events since December 2025, letting Cloudflare disambiguate the 01-02 anomaly (narrative: attack tied to U.S. military action in Venezuela) from chronic loose- export-policy behaviour. The per-event page (leak-462460) names leaker (AS8048), prefixes, and neighbor ASes. BGPKIT monocle complements Radar with per-adjacency directional evidence from route-collector data.
  • sources/2025-07-16-cloudflare-1111-incident-on-july-14-2025 — Radar auto-flagged Tata Communications (AS4755) advertising 1.1.1.0/24 at 21:54 UTC during the outage; hijack was real but not the cause — it became visible only because Cloudflare had withdrawn its own advertisement. Radar provides the receipts both for "this is / isn't the cause" questions.
  • sources/2026-04-17-cloudflare-agents-week-network-performance-update — Radar's Internet Quality surface as the data substrate for Cloudflare's periodic network-performance update posts. The 40 % → 60 % "fastest in top 1,000 networks" shift (Sept → Dec 2025) and the 6 ms average gap to next-fastest provider are both computed from RUM-via-error-page comparative measurements aggregated with trimean of connection time against CloudFront / Google / Fastly / Akamai. First wiki instance of Radar as comparative-performance substrate (distinct from the BGP-forensics role).
  • sources/2026-04-17-cloudflare-introducing-the-agent-readiness-score-is-your-site-agent-ready — Radar's Adoption of AI agent standards surface on AI Insights — weekly scan of the top 200 k filtered domains for presence of each agent-readiness check. First wiki instance of Radar as standard-adoption-measurement substrate. Quantifies the current baseline: robots.txt 78 % / Content Signals 4 % / markdown content negotiation 3.9 % / MCP Server Cards + API Catalog combined < 15 sites in the entire dataset. Turns isitagentready from a diagnostic tool into a public adoption dashboard with a feedback loop (sites fix → next scan measures the delta).
  • sources/2026-04-17-cloudflare-redirects-for-ai-training-enforces-canonical-content — Radar's Response Status Code Analysis surface on AI Insights (and per-bot bot-directory pages) — aggregate + per-bot 2xx / 3xx / 4xx / 5xx distributions filterable by industry and crawl-purpose (Training / Search / Assistant). Aggregate representative: ~74 / 13.7 / 11.3 / 1.2 %; GPTBot representative: ~83 / 10 / 5.1 / 2.2 %. First wiki instance of Radar as crawler-response-policy measurement substrate — the 3xx slice is the visible adoption curve for Redirects for AI Training across the web; the 403 / 402 slices are visible adoption of WAF hard-block and pay-per-crawl policies.
Last updated · 200 distilled / 1,178 read