We restore-tested the backups end to end
A backup is worth what has been restored from it — so we did it. From the off-site backup (every 6 hours, to EU storage) we restored the entire database into a separate instance and counted: every table, every row and every user account came back intact. We checked the canonical shift-log PDFs separately: every one of them has its backed-up file, byte-for-byte identical to the original. The data itself restores in seconds. The full recovery time — which includes rebuilding a machine — is still being measured, and we will only state it once we can back it with a number.
- Every table and every user account came back intact — counted, not assumed
- Every canonical shift-log PDF is in the backup, byte-identical
- What we have not measured, we say so — the full recovery time is not established yet