Overview / Resilience · B04
Standards we keep, outcomes you can check
Delivery standards for every engagement, written so you can hold us to them.
No uptime figures or certifications are claimed. Each standard below produces evidence you can inspect.
| Standard | What we do | Evidence you receive |
|---|---|---|
| Separated copies | Backups held apart from primary data, with one copy protected from ordinary deletion | Backup configuration summary in environment record |
| Tested restores | Restore tests on an agreed schedule, including partial and full restores | Restore test notes with time taken and issues found |
| Traceable runs | Every processing run logged: inputs, outputs, rejections, duration | Run log available on request or in periodic reports |
| Access hygiene | Named accounts, least privilege, scheduled reviews, prompt removal | Access list with review dates |
| Change records | Changes to environments and jobs recorded with reason and date | Change history |
| Handover-ready | Runbooks written for a competent stranger | Runbooks, checked at handover |
| Clean exit | Documented export and account handover | Exit plan |
When things break
They do. A source changes format, a provider has an incident, a credential expires. Resilience is about the second hour, not the first minute: knowing what failed, what was affected and how to recover.
For ongoing work, each runbook defines what counts as a failure, who is notified, and the recovery steps. During working hours we act on failures directly. Outside hours, arrangements are designed so that a failed run stops cleanly and waits, rather than producing half-finished output.
After any significant incident we write a short, blame-free note: what happened, impact, fix, and what changes.