Wie genau sind wir?
Die meisten Ausfall-Tracker veröffentlichen ihre eigene Genauigkeit nicht. Wir schon. Diese Zahlen werden live aus unserer Monitoring-Datenbank berechnet und aktualisieren sich, während das System läuft.
Anbieter-Übereinstimmung
0%
Wie oft unser Urteil mit der offiziellen Statusseite übereinstimmt (letzte 30 Min.)
Ø Lösungsdauer
0 min
Mittlere Vorfallsdauer in den letzten 30 Tagen
Probe-Erfolgsrate
100.00%
622,512 Checks in 24h
Monitoring-Regionen
4
Multi-Region-Konsens für Ausfall-Urteile erforderlich
Überwachte Dienste
680
680 im Katalog
Mit Anbieter-Status
144
Mit offiziellen Statusseiten abgeglichen
Aktive Vorfälle
0
0 in 24h geöffnet
Community-Meldungen
0
Nutzerbeiträge in 24h
Was unsere Urteile genau macht
- Multi-region consensus. No single probe ever decides an outage. A failure in one region with others healthy is treated as probe noise.
- Persistence gate. An incident is only opened if the same region fails across ≥2 consecutive rounds — kills false positives from transient blips.
- Vendor status ingestion. We pull the official status page (Statuspage.io, Slack, Heroku) every 2 minutes and weight it as the highest-trust signal.
- Body fingerprinting. A 200-OK maintenance page is not "up." We verify expected HTML markers and detect known failure patterns.
- Community signal engine. User reports are baseline-compared (5× normal volume + ≥3 countries triggers escalation), with anti-spam scoring.
- Verdict explainability. Every service page shows exactly which signals drove the current status. No black box.
Ehrliche Grenzen
- Our probes run from cloud datacenters. Some vendors rate-limit those IPs, which can show as latency we'd never see from a real user. We compensate with retries and baseline learning.
- We can't see inside a service — payment failures, login outages, regional CDN issues for end-user IPs are partly inferred from community reports.
- Status page agreement is a quality signal, not ground truth. Vendors sometimes report green during real outages.
Öffentliche API
Jeder kann unser Urteil für jeden überwachten Dienst abrufen — keine Auth, CORS offen, 30s Edge-Cache.
GET /api/public/v1/status/<slug>{
"service": { "slug": "...", "name": "...", "domain": "..." },
"status": "online | slow | degraded | partial_outage | down",
"confidence": { "confidence": 0-100, "message": "..." },
"health": { "score": 0-100, "label": "Excellent | ..." },
"dns": { "ok": true },
"tls": { "expires_at": "ISO" },
"vendor": { "status": "...", "checked_at": "ISO" },
"verdict": { "reasons": [ ... ] }
}