Our infrastructure is 3x redundant, self-checking, self-healing and distributed over several availability zones to ensure high availability.
We use a Kubernetes cluster, that:
- stretches over two different Google Cloud availability zones
- runs every production deployment on at least one instance per zone
- features a synchronized master in each zone.
Our cluster has automatic health checks on both hard- and software. This ensures an appropriate distribution of load over all healthy machines and automatic discovery, isolation, and replacement of faulty components resulting in a very high reliability.
Our database servers have a standby copy in a second availability zone with automatic handover in the case of catastrophic failure and a separate backup mechanism allowing us to go back to any point in time within the last 7 days.