About

isitdown.io exists to give you the fastest, most accurate data on any service experiencing an issue — whether that's a website, an API, a Minecraft or game server, a publisher's status feed, or the cloud provider underneath all of them.

Most "is X down?" tools answer the wrong question. They tell you that one service is down. We tell you what that means — whether the outage is global or regional, whether it's confirmed by multiple monitoring regions or one bad blip, whether it's the service itself or the cloud provider underneath, and whether other services you depend on are likely to be affected too.

What we check

Different layers of the internet need different probes. We run all of them:

• Websites + APIs — direct HTTP/HTTPS probes from US East, US West, Europe, and Asia in parallel. Status code, latency, TLS handshake, redirect chain, response headers — captured per region. Example: Discord.
• Minecraft servers — native Server List Ping (SLP) handshake. Real player count, version, MOTD, ping latency — Java + Bedrock, BungeeCord/Velocity proxies aware. All tracked Minecraft servers →
• Game servers (Source engine) — A2S_INFO UDP queries for CS2, TF2, Garry's Mod, Rust, Ark community servers. Map, players, server name. All tracked game servers →
• Publisher status feeds — direct polling of the official status pages for Discord, GitHub, Twitch, Cloudflare, Reddit, Epic Games. We surface the publisher's own classification, not crowd reports.
• Diagnostic tools — SSL certificate inspection (/ssl), DNS records (/dns), redirect chains (/redirect), security-header grading (/headers) for any host you give us.
• Cloud infrastructure cascades — when AWS or another major provider has an incident, /infra shows you exactly which consumer services we monitor publicly depend on it. The dependency map nobody else surfaces.

How "fastest" actually works

Speed comes from three places:

• Direct probes, not crowd reports. When you ask if Discord is down, we run a live HTTP request to discord.com from four regions in parallel and answer in 200-800ms. We don't aggregate user reports from social media — those are noisy and slow to react.
• Background monitoring. The 100+ most-checked services are probed every 5 minutes so a check-page hit serves cached fresh data instantly. Less common targets get a live check on first request, then cached for 60 seconds.
• Debounced incident state machine. A single failed probe doesn't trigger an outage report. Two consecutive failures across the cycle do — and recovery requires two consecutive successes. The result: zero false-alarm cascades on transient blips, but legitimate outages surface within ~10 minutes of the second failure.

How "most accurate" actually works

We're explicit about what we know and what we don't. If our monitoring infrastructure can't reach upstream targets (sandbox / restricted egress / a rare upstream-of-us internet event), the homepage swaps the red outage strip for an amber "MONITORING OFFLINE" banner — we don't falsely accuse 100 healthy services of being down. If a service's status is "unknown" because its first probe hasn't completed, we say "PENDING" with a grey dot, not green.

Same standard for the cloud cascade tracker: we only tag a service as AWS-dependent (or Cloudflare, or GCP) when the dependency is publicly stated — status pages, official tech-stack pages, AWS case studies, BGP/IP-range data. Wrong dependencies during a real outage are worse than no dependencies; we'd rather show fewer we're certain of than more half-guessed.

What you can do with the data

• Subscribe to alerts — webhook or email, per-service, anonymous. API docs at /api.
• Embed status badges — every check page has a Copy-snippet block. The SVG updates automatically with the live state.
• Read the historical record — 30-day uptime calendar on each website check page, RSS feed of active outages at /feed/outages.xml.
• Watch the cascade — when a major cloud provider has an incident, /infra shows the live blast radius across our catalog.

What we don't do

No accounts. No paywall. No tracking the user across sessions. No editorial spin on whether an outage is "newsworthy." No charging operators to remove their service from our list. The product is the data; the data is free.

Contact

Questions, corrections, or want a service added to the catalog? Visit our contact page. We're especially happy to receive corrections to the cloud-dependency tags — being wrong there matters more than anywhere else.