Resilient software systems

Failure is the normal case

Disks die, zones go down, networks partition. A resilient system plans for that: state is replicated, requests are processed idempotently, and a failover runs without anyone touching a keyboard. We have built such systems since 2015 for financial services, utilities and manufacturing.

How we work

1. Two weeks of analysis: which failures must be survived, which data loss is acceptable?
2. Target architecture with replication, queueing and recovery, agreed with your team.
3. Reference implementation in Go with load and failover tests in your environment.
4. Hand-over with operations manual and training.

Outcome

A system whose failure behaviour is documented and tested. You know what happens when a site drops out because we tried it together.