Measure recovery time and recoverable data loss separately
Management asks for recovery objectives, but a single successful local restore is being treated as a guarantee.
- Focused work estimate
- 5h + prerequisites
- Priority in the scenario
- High
- Engineering practice
- Reliability planning · Experimental design
Estimated field mix
- Site reliability60%
- Performance engineering40%
Field percentages are editorial estimates of the ticket's engineering focus. They total 100%; they are not measured time, proficiency scores, or ownership evidence.
Review it, then add it to your workspace.
The board opens an editable draft; nothing is saved until you confirm it. Sign-in and workspace permissions apply, and Demo boards remain ephemeral.
Project context
A fictional reporting service has successful backup jobs but no recent restore rehearsal. Database rows reference objects with different retention schedules.
Setup prerequisites
- Create disposable local database and object-store fixtures with synthetic reports; author backup and corruption examples.
Preceding work
Complete these dependencies, or supply their agreed outputs before taking this ticket.
- BRESTORE-101 · Inventory the data required to restore one report
- BRESTORE-102 · Define a consistent backup manifest across stores
- BRESTORE-103 · Guard restore commands against non-scratch targets
- BRESTORE-104 · Verify backup bytes before applying restored data
- BRESTORE-105 · Restore database references before enabling object reads
- BRESTORE-106 · Rebuild derived indexes from restored authoritative records
- BRESTORE-107 · Check application compatibility against restored schema versions
Acceptance criteria
- Measure each restore phase under stated dataset and machine limits.
- Compute the gap between backup boundary and failure time.
- Report untested scale and dependency assumptions.
Implementation constraints
- Local rehearsal establishes observations, not production RTO or RPO guarantees.
Verification to include
- Measure a complete synthetic recovery.
- Expose an unrecoverable post-backup write explicitly.
Deliverables
- Recovery assessment.
Rollout and recovery
Use results to propose objectives; repeat in the target environment before committing to them.
Value of the work
For the engineer: Practice recovery consistency, integrity validation, and honest recovery objectives.
For the team: Replace backup-job optimism with a repeatable restoration assessment.
Evidence boundaries
Outcome Evidence: Tests, patches, and runbooks are requested deliverables. They become Outcome Evidence only through a qualified Mission and immutable Evidence IDs.
Ownership Evidence: Independent adaptation must be observed under a declared verification policy and cite immutable Evidence IDs. Completing a planning ticket establishes no Ownership Evidence.