I85PREP
⌘ K
HPE ASE Storage Architect study materials
HPE ASE Storage Architect study resources

HPE ASE Storage Architect: Storage Solutions Design and Architecture

HPE ASE Storage Architect · Work through your exam, one topic at a time.

All guide sections

Storage Solutions Design and Architecture

This domain tests whether you can convert a customer's workload and availability requirements into a defensible HPE array design: controller platform, RAID layout, port counts, capacity headroom and data-service placement. Expect scenario questions where two answers are technically possible and only one matches the stated SLO.

Requirements before products

  • Workload profile: IOPS, read/write mix, block size, random vs sequential, latency SLO and peak-to-average ratio.
  • Capacity: usable today, three-year growth, thin overprovisioning, snapshot and replication reserve kept separate from user data.
  • Availability: RPO/RTO, site distance, stretched cluster or standalone DR.
  • Connectivity: FC 32/64 Gb, iSCSI 25/100 GbE or NVMe/TCP; host count and OS mix drive port fan-in.

Platform behaviour that changes the arithmetic

  • Primera and Alletra 9000 use a multi-controller mesh-active design with ASIC-assisted data path and RAID 5/6 with distributed sparing, so there are no dedicated hot-spare drives; RAID 6 8+2 is roughly 80% efficient.
  • Alletra MP B10000 is disaggregated: compute nodes plus NVMe JBOF shelves, scaled over NVMe/TCP, so capacity and performance scale independently.
  • Alletra 6000 (Nimble) uses CASL with triple-parity RAID in scale-out clusters; MSA is an entry controller pair with classic RAID 1/5/6/10.

Sizing arithmetic

  • Usable = raw x RAID efficiency, minus metadata and snapshot reserve.
  • Front-end bandwidth = IOPS x block size; plan for 50-60% of line rate after protocol overhead.
  • Backend SSD IOPS per drive and rebuild time matter as much as capacity.

Worked example

Requirement: 150 TB usable today, 20% growth (180 TB), under 1 ms latency at 30,000 IOPS of 8 KB with a 70/30 read/write mix. Choose RAID 6 8+2 on 3.84 TB SSDs. Raw needed is 180 / 0.8 = 225 TB, so 225 / 3.84 = 59 drives, rounded to 64 drives (eight 8-drive RAID 6 sets). Usable becomes 64 x 3.84 x 0.8 = 196 TB, leaving about 16 TB for snapshots and metadata. Bandwidth is 30,000 x 8 KB = 240 MB/s, comfortably within two 32 Gb FC ports per fabric at 50% utilisation, but present four ports per fabric for N+1 and host fan-in.

Traps that cost marks

Mixing decimal TB with TiB; forgetting snapshot and Remote Copy reserve; sizing to average rather than peak; assuming NVMe-oF removes the need for fabric planning; choosing RAID 6 on very small drive counts; ignoring per-shelf power and rack density; and designing a single fabric for simplicity.