PATTERN Cited by 1 source
Snapshot-clone rapid failover¶
Intent¶
Achieve sub-minute volume-level DR recovery by pre-provisioning a copy-on-write clone from the latest replication snapshot in the DR region, so failover becomes a traffic redirect rather than a storage rebuild.
Mechanism¶
- Continuous replication (e.g. SnapMirror) delivers snapshots to the DR region at a fixed interval
- Scheduled automation (e.g. daily) creates a copy-on-write clone (e.g. FlexClone) from the latest available snapshot
- The clone is presented to a pre-staged compute instance (read-only mode)
- On DR trigger, application traffic is redirected to the pre-provisioned instance — no storage provisioning needed at failover time
The clone creation is sub-2-minute (metadata-only operation), and because it's done proactively, the actual failover path is even faster — it's purely an application-layer cutover.
Key properties¶
- Near-zero additional storage: clone shares blocks with parent via CoW
- Non-disruptive: clone creation doesn't interrupt ongoing replication
- Pre-provisioned: eliminates storage-provisioning latency from the failover critical path
- Bounded RPO: determined by the replication schedule (not the clone refresh schedule)
- Bounded staleness of read-only view: determined by the clone refresh schedule (e.g. daily)
Trade-offs¶
| Advantage | Cost |
|---|---|
| Sub-15-minute read-only recovery | Read-only view may be up to clone-refresh-interval stale |
| Minimal storage overhead | Full read-write still requires SnapMirror break sequence |
| Replication not interrupted by failover | Requires orchestration automation for clone refresh |
| Works with any snapshot-capable storage | Specific to storage systems with CoW clone support |
Production reference¶
S&P Global's Capital IQ platform on Amazon FSx for NetApp ONTAP: SnapMirror at 15-minute intervals + daily FlexClone refresh achieves sub-15-minute read-only failover for a global financial data platform serving strict RTO/RPO requirements.
(Source: sources/2026-07-07-aws-sp-globals-disaster-recovery-fsx-netapp-ontap)
DR tier mapping¶
This pattern implements a hybrid warm-standby (read-only) + pilot-light (read-write) shape: - The pre-provisioned read-only clone is warm standby (ready to serve immediately) - Full read-write requires the patterns/two-phase-dr-read-then-write orchestration (pilot-light activation)
Related¶
- patterns/two-phase-dr-read-then-write — the companion pattern for full read-write recovery
- patterns/pilot-light-deployment — the DR tier the full read-write phase maps to
- patterns/warm-standby-deployment — the DR tier the read-only phase maps to
- concepts/flexclone-instant-volume-copy — the CoW clone mechanism
- concepts/snapmirror-replication — the replication mechanism
- concepts/disaster-recovery-tiers — where this sits on the ladder