Skip to content

3. Data Protection and Recovery

Separate availability, durability and recoverability, then choose protection mechanisms using failure domains, recovery point and recovery time.

Level: Foundation · Time: 45 minutes · Prerequisite: Storage Building Blocks

MechanismPrimary purposeDoes not replace
RAID/mirroringremain available through selected device failuresindependent backup
Snapshotpreserve a point-in-time logical versionisolated off-system copy
Replicationmaintain another copy, synchronously or asynchronouslyversioned recovery from every corruption/operator error
Erasure codingtolerate fragment loss with efficient capacitycorrect placement and repair operations
Backuprecover independent historical copiescontinuously available primary storage

RPO bounds acceptable data loss in time. RTO bounds acceptable restoration time. Neither is proven by configuration alone; recovery must be tested.

Use the Universal RAID Planner to compare RAID 6, RAID 10 and RAID-Z2. Then use the Erasure Coding Lab to fail fragments at and beyond tolerance.

  1. Why is a snapshot on the same failed array not an independent backup?
  2. What can make a mathematically recoverable erasure-code layout operationally unsafe?
  3. When might RAID 10 be preferred over a parity layout despite lower usable capacity?

Capacity formulas do not predict rebuild time, correlated failure, latent corruption or workload impact. Validate vendor implementation and measured recovery behavior.

Continue to SAN Architecture.