Reliability & APIs Core Web Vitals Field Data Measures What Real Visitors Experience "Core Web Vitals report LCP, INP and CLS from real Chrome page loads at the 75th percentile, and the thresholds only mean something once you know which segment and which percentile you are r
Reliability & APIs Cloudflare Measures the Markdown It Serves to AI Agents "Cloudflare's Markdown for Agents reached general availability with analytics, logs, a playground, a GraphQL dataset and a tracing span, and the docs say what those numbers can and cannot pr
Reliability & APIs QuePaxa Lets a Global Control Plane Keep Deciding While Links Stall Cloudflare runs Meerkat, an experimental consensus service, on QuePaxa, an algorithm with an optional leader and no fault detector on the decision path.
Reliability & APIs Grafana Puts a Sampling Budget Behind Every Trace It Keeps Grafana's Adaptive Telemetry spreads one sampling budget across services, keeps a trace per attribute set, and holds dropped traces by ID for 24 hours.
Reliability & APIs Cloudflare Reclaims 100 TB of RAM by Rebuilding Its Hash Ring Cloudflare's load-balancing service kept 100,000 hash points per server to spread cacheable requests, and the structures behind them reached 6 GB in some processes. A rebuild that cut those
Reliability & APIs One Power Event in a London Zone Took Down 41 Google Cloud Products A power event inside europe-west2-b interrupted a reported 41 Google Cloud products for 7 hours 52 minutes on 20 October 2026, and the rest of the London region saw elevated errors alongside it. Revolu
Reliability & APIs Designing Alerts That Explain an Incident An alert is a message that arrives on a phone, and design decides whether it explains an incident or adds to the noise. The burn-rate windows in Google's SRE Workbook and the grouping, inhib
Reliability & APIs Cloudflare Hyperdrive Serves MySQL From One Canonical Copy of a Query Result Cloudflare took Hyperdrive's MySQL support to general availability on 14 October 2026 after 20 fixes, including a cache bug that made roughly 0.001% of queries return empty results, and repo
Reliability & APIs Making Retries Safe: A Practical Guide to Idempotency Keys A practical explanation de idempotency keys apis.
Reliability & APIs API Rate Limits Are Part of AI Agent Architecture An AI agent that ignores rate limits is an uncontrolled client with a retry loop.
Reliability & APIs OpenTelemetry Standard vs Vendor Observability Boundaries OpenTelemetry standardizes how telemetry is produced and moved, but not what a vendor backend does after ingestion.