Restore verification

Restore verification ledger

Every scheduled AWS Backup restore test, verified by an agent that fingerprints the restored database and compares it with the previous restore. Times are UTC. The ledger stores counts and salted digests, never data.

2restore tests verified
2 of 2passed
12 min 35 smedian restore time
2026-11-13last verified restore
2 of 2recovery objectives met
CompletedDatabaseVerdictRestore timeRTORPOSummary
2026-11-13 14:28elastic-hackaton-dbSUCCESSFUL16 min 39 smet
16 min of 30 min
met
3.1 h of 24 h
The MySQL 8.4.13 restore contains all four expected business tables—customers (20 rows), inventory (50), orders (29), and order_items (58), totaling 157 rows—with unchanged schemas and no missing tables. Compared with the previous verified restore, customers, orders, and order_items grew by 15, 24, and 53 rows respectively, and the carried-over append-only history checks confirmed all five prior rows in each table are intact. The newest created_at in those three tables is 2026-11-13T13:06:01Z, 50 minutes 32 seconds before the backup; inventory stayed at 50 rows with its original newest created_at, which is plausible for a mutable stock table. Every table was readable and fingerprinted, and both declared foreign keys plus order_items.sku to inventory.sku resolved with zero orphaned references. RTO was met at 16 minutes against a 30-minute target (999-second restore), and RPO was met at 184 minutes against a 1,440-minute target; no data-integrity failure was found.
2026-11-13 12:44elastic-hackaton-dbSUCCESSFUL8 min 31 sThe MySQL 8.4.13 restore contains all four business tables: customers (5 rows), inventory (50), orders (5), and order_items (5), totaling 65 rows. Every table was readable and fingerprinted, and each newest created_at timestamp is 2026-11-13T10:50:12Z, 89 seconds before the 2026-11-13T10:51:41.852Z backup. Both declared foreign keys and the inferred order_items.sku to inventory.sku reference were checked, with zero orphaned rows across all three relationships. This is the first recorded restore test, so no prior fingerprints or row counts are available for historical-drift comparison. On readability, freshness, table presence, and referential integrity, the restore is usable and complete.