Snapshot restore → Point-in-time restore

Status: Complete (via workaround)
Priority: Must have
Audience: SRE, security, platform

Terminology

Was	Now
Snapshot restore	Point-in-time restore

The recovery capability for Serverless is point-in-time restore. Older conversations and runbooks may still say “snapshot restore” — treat those as the same requirement under the updated product language.

Why it matters

Production Serverless search needs a disaster-recovery path that meets your RPO and RTO. Waiting to restore an entire historical corpus in the cloud should not be a prerequisite — focus recovery on the interactive Serverless dataset.

Current state

Accepted as complete via workaround for GA. Keep the interim restore runbook current and align RPO/RTO for the interactive Serverless dataset.

Recommended approach

Define RPO/RTO for the interactive Serverless tip (user-facing search), not the full historical estate.
Keep the accepted workaround / interim restore procedure documented.
Document restore ownership and verification queries in the runbook.
Revisit native point-in-time restore when Product ships further improvements.

Acceptance criteria

Accepted via workaround for GA
Agreed RPO/RTO for interactive Serverless indices documented
Restore runbook covers who triggers restore and how search is verified after