An administrator manages a group of HPE Alletra MP B10000 arrays through the DSCC console.

The HPE Alletra MP B10000 (Block) utilizes a disaggregated shared-everything architecture where capacity is distributed across multiple NVMe drives and enclosures. To ensure 100% data availability, the system allows administrators to define the level of resilience required via the High Availability (HA) settings.

Architecturally, there is a direct trade-off between the level of hardware resilience and the achievable space efficiency (usable capacity).

Enclosure Level HA (Option D): This is the most resilient setting. It ensures the system can survive the total failure of an entire drive enclosure (JBOF) without losing data. To achieve this, the system must distribute parity and data stripes across different enclosures. This "vertical" redundancy requires a larger percentage of raw capacity to be reserved for parity, thereby reducing the net space efficiency.

Drive Level HA (Option A): This setting protects against individual drive failures (similar to traditional RAID 6 or RAID-TP) but assumes the enclosure itself remains operational. Because the stripes can be optimized more densely within fewer hardware boundaries, the system requires less "overhead" capacity to maintain the protection state.

By changing the High Availability option to Drive Level, the administrator instructs the Alletra MP software to prioritize usable capacity over enclosure-level fault tolerance. This is a common optimization for customers who have multi-enclosure systems but prefer to maximize their ROI on raw NVMe flash. It is important to note that changing this setting may require a re-striping of existing data and should be done in accordance with the customer’s risk profile and SLA requirements. The sparing algorithms (Options B and C) manage how much space is set aside for automatic rebuilds, but the primary driver of bulk space efficiency in a multi-enclosure MP cluster is the HA policy selection.