We do not publish uptime figures
We do not yet run availability monitoring, so there is no measured record to show, and we would rather say that than publish a number we cannot stand behind. If we start measuring it, this page is where the figures will appear.
Incidents that affect customers are reported here and emailed to affected organisations. If something is wrong right now, tell us. That is still the fastest signal we have.
No availability SLA
The platform is built for prototyping, validating and proving applications, not for running systems that require high availability. Applications that need that should graduate to a platform built for it, and we help with that move.
One region
Everything runs in eu-central-1 (Frankfurt), on a dedicated instance per environment. Today that is the only region we operate.
No zero-downtime deploy
Every deploy briefly interrupts the running application. That includes a rollback: since #548 a deploy can be pinned to an earlier commit, so recovering from a bad deploy is redeploying the last good one, but it is a redeployment, not a hot restore.