HCIE-Storage cheat sheet

Storage (HCIE-Storage Expert Track)

• RAID capacity rule: usable capacity = N × smallest member disk; RAID 5 tolerates 1 failure, RAID 6 tolerates 2, RAID 10 tolerates 1 per mirrored pair and requires even disk count.

• RAID 5 write penalty = 4 I/Os per host write (read old data, read old parity, write new data, write new parity); RAID 6 penalty = 6; RAID 10 penalty = 2.

• Rebuild time ≈ array capacity ÷ rebuild rate; a 10 TB member at 100 MB/s takes roughly 28 hours, so larger single members increase exposure window.

• Thin provisioning over-subscription ratio = sum of provisioned LUN capacity ÷ physical pool capacity; alarm thresholds typically trigger auto-expansion or write rejection at 100%.

• Snapshot copy-on-write consumes space only for changed blocks; copy-on-first-write redirects the first write of each block, so space cost tracks write churn, not elapsed time.

• Clone (full copy) is independent of the source after creation; snapshot depends on the source LUN, and deleting the source invalidates dependent snapshots unless promoted.

• Cache hit ratio = read hits ÷ total reads; write-back cache requires battery or capacitor protection, and falls back to write-through when protection fails.

• Queue depth and IOPS relationship: IOPS = queue depth ÷ latency; at 1 ms latency, one outstanding I/O yields about 1,000 IOPS, so concurrency drives small-block performance.

• Multipath policy choices: failover (one active path), round-robin (per-I/O distribution), and least-queue-depth; ALUA arrays report asymmetric access states per target port.