Worst of the cards beside it; their median.
Operationalmedian 100.00% uptime
Signing in, managing environments
Operational100.00% uptime
The control panel API
Operational100.00% uptime
Assets served from the edge, not fetched from us
Operational100.00% uptime
Authoritative DNS for customer domains
Operational100.00% uptime
Off-site copies of customer data and of this service
Operational100.00% uptime
Build artifacts and container images
Operational100.00% uptime
Billing and mail, which are somebody else's to run
OperationalProvider reports operational
Customer environments in Falkenstein 01
Customer environments in Helsinki 02
Customer environments in Zagreb 01
Customer environments in New York 01
Reported by the providers underneath us, and shown here because it may explain what you are seeing. A provider's status page is updated once an incident is understood, so it can lag what we are measuring directly.
Announced work, ours and our providers'. Downtime inside an announced window is not counted against an availability commitment, which is why the window is published before the work rather than after it.