I85PREP
⌘ K
HCIE-Storage study materials
HCIE-Storage study resources

HCIE-Storage: Data Protection and Disaster Recovery

HCIE-Storage · Work through your exam, one topic at a time.

All guide sections

Data Protection and Disaster Recovery

Domain 4 - Data Protection and Disaster Recovery

This domain tests mapping RPO and RTO to the correct OceanStor feature and configuring it so the objective is actually met: HyperSnap, HyperClone, HyperCopy, HyperReplication, HyperMetro, and backup to OceanProtect or NDMP tape.

Matching technology to objective

  • HyperSnap: point-in-time copy in seconds, RPO in minutes. Snapshots live in the source pool, so they do not survive LUN or array loss.
  • HyperClone: full copy split from the source; independent and space-consuming.
  • HyperReplication: remote replication, synchronous for RPO = 0 within the latency budget, asynchronous over longer distances with a measurable lag.
  • HyperMetro: active-active dual-write, RPO = 0 and near-zero RTO; both arrays serve I/O and a quorum server arbitrates.
  • Backup: a separate media and failure domain; only backup protects against ransomware and logical corruption.

HyperMetro rules

Both LUNs must be the same type and size, arrays must run compatible OceanStor OS versions, and inter-site round-trip latency must fit the synchronous budget, typically 10 ms RTT or less. The quorum server belongs in a third failure domain with independent power and network. If the inter-array link and the quorum server are both unreachable, the pair cannot arbitrate and I/O is suspended on both sites to prevent split-brain.

Worked example

Two data centres 40 km apart need RPO = 0 for a database. Option A is HyperMetro with dual-write and a quorum server at a third site. Option B is synchronous HyperReplication with manual failover: RPO = 0, but RTO is minutes to hours. If near-zero RTO is also required, only HyperMetro qualifies. If the latency budget cannot be met, the RPO requirement must be renegotiated to asynchronous replication.

Traps

Treating snapshots as backups; placing the quorum server inside either data centre; ignoring that clone and snapshot space comes from the production pool; failing over without reversing replication; leaving RTO untested, so the recovery plan exists only on paper.