HPE ASE Storage Architect key terms

A disk layout that stripes data with two independent parity blocks, so the array keeps running after two drive failures. It costs two drives of usable capacity but suits large nearline disks where rebuild times are long.

A layout that first mirrors pairs of drives, then stripes across the mirrored pairs. It offers fast rebuilds and high write throughput but uses only half the raw capacity, making it common for latency-sensitive workloads.

A protection method that splits data into fragments and computes parity fragments so the original survives loss of a defined number of pieces. It delivers durability with far less overhead than mirroring, at the cost of CPU and rebuild work.

Presenting a logical volume larger than the physical space currently reserved for it, allocating capacity only as data is written. It improves utilization but requires monitoring and policies so pools do not fill unexpectedly.

Reserving the full physical capacity of a volume at creation time, so writes never fail from pool exhaustion. It wastes space when volumes stay partly empty but gives predictable performance and simpler capacity planning.

A collection of physical drives or backend capacity from which logical volumes are carved. Pool-level policies such as RAID level, tiering, and deduplication apply to every volume created inside it.

An identifier for a logical volume presented to a host over a storage protocol. The host sees it as a disk device, and masking or mapping rules decide which initiators may access it.

A control that restricts which host initiators can see and access a given logical unit. It prevents unintended hosts from writing to a volume and is a basic building block of SAN access control.

On a Fibre Channel fabric, a set of rules grouping ports or worldwide names so members can communicate. Correct zoning limits disruption from misbehaving hosts and reduces the paths a server can discover.

A globally unique 64-bit identifier burned into a Fibre Channel port or node. Administrators use these names in zoning, masking, and host registration instead of physical port numbers.

The process by which a node port registers with the fabric switch to receive its address and learn fabric parameters. Without a successful login the host cannot be zoned or see storage targets.

A protocol that carries SCSI commands over TCP/IP, letting hosts use Ethernet networks to reach block storage. It is cheaper than Fibre Channel but depends on network design, jumbo frames, and authentication.

A challenge-response method that proves an initiator or target knows a shared secret without sending it in clear text. It is commonly configured on iSCSI sessions to block unauthorized hosts from logging in.

A standard that encapsulates Fibre Channel frames inside Ethernet so storage and general traffic share converged switches. It needs lossless Ethernet configuration such as priority flow control to avoid dropped storage frames.

Host software that presents several physical paths to one volume as a single device, load balancing across them and failing over when a path breaks. It prevents a single adapter or switch loss from taking storage offline.

A transport that extends the low-latency NVMe command set across a network such as Fibre Channel or RDMA Ethernet. It removes much of the legacy SCSI stack overhead and is chosen for latency-sensitive workloads.

A point-in-time logical copy of a volume that usually shares unchanged blocks with the source. It enables fast rollback and backup without duplicating the full data set, but consumes pool space as changes accumulate.

A fully independent copy of a volume or snapshot that can be read and written without affecting the source. Clones are often used for test and development, and they consume capacity for any changed blocks.

Copying data from one array or site to another to support failover or read distribution. Synchronous mode keeps both copies identical but adds latency, while asynchronous mode tolerates distance at the cost of possible data loss.

The maximum acceptable amount of data loss measured in time before a disruption. It drives choices such as replication mode and snapshot frequency, because a tighter objective demands more frequent copying.

The target duration within which a service must be restored after an outage. It shapes architecture decisions such as automatic failover, redundant paths, and standby systems rather than backup frequency alone.

A stretched-cluster arrangement in which hosts automatically move I/O to a remote array when the local one fails, keeping the same presented volumes. It delivers transparent failover but requires low-latency links between sites.

The period during which backup jobs must complete without harming production performance. It constrains job sizing and scheduling, and is often shortened by snapshot-based or incremental-forever methods.

A job that copies only data changed since a chosen reference point, reducing time and network use. Restores may need the chain of prior jobs unless the method periodically synthesizes a full image.

A retained copy that cannot be altered or deleted, even by an administrator, until its retention period expires. It defends against ransomware and insider deletion by making recovery points tamper-evident.

A deliberate separation that prevents direct network access between production and a protected copy, such as an offline vault or one-way link. It blocks remote attackers from encrypting or deleting the recovery data.

Rules that define how long copies are kept, how often they are taken, and when they expire. It balances compliance and recovery needs against storage cost and must align with the stated recovery objectives.

Detecting and storing only one copy of repeated data blocks, replacing duplicates with references. It greatly reduces capacity for similar data such as virtual machines, but consumes CPU and memory and can slow restores.

Encoding data so it occupies fewer bytes on disk, usually applied inline or after writing. It lowers capacity use with modest CPU cost and combines with deduplication, though already-compressed or encrypted data may not shrink.

Releasing pool space that is no longer referenced, such as after deleting files, snapshots, or virtual machines. Without reclamation, thin volumes stay inflated and arrays report free space that cannot actually be reused.

Controls that cap or guarantee IOPS and bandwidth per volume, host, or application. They stop one noisy workload from starving others and make performance predictable in shared arrays.

A recorded set of normal metrics such as latency, IOPS, and queue depth captured while the system is healthy. Comparing live data to it makes anomalies visible before users report problems.

Fast memory or flash that holds recently or frequently used data to reduce read and write latency. Cache sizing and write-back policy strongly affect performance, and losing unprotected cache can risk data.

Automatically moving data between fast and slow media based on access patterns. It lowers cost while keeping hot data responsive, but requires monitoring so sub-LUN movement does not cause unexpected load.

An idle drive kept available to automatically replace a failed member of a RAID group. It shortens exposure to a degraded array, though rebuild time and spare size must match the drives being protected.

The period needed to restore redundancy after a drive failure by reconstructing data onto a replacement. Larger drives and heavier production load lengthen it, raising the chance of a second failure during recovery.

Storing data in encrypted form on media so that removing a drive does not expose its contents. Keys are managed by the array or an external manager, and losing them makes the data unrecoverable.

Protecting data while it crosses a network between host and storage, typically with TLS or IPsec. It closes the gap left by media encryption when links pass through shared or untrusted networks.

The generation, storage, rotation, and escrow of cryptographic keys used for storage encryption. Poor key handling can lock an organization out of its own data, so separation of duties and backups of keys matter.

Granting management permissions according to job roles rather than to individuals. It limits who can delete volumes, change replication, or alter retention, and supports audit requirements for separation of duties.

A chronological record of administrative and access events on the storage system. It supports investigations and compliance evidence, and should be shipped off the array so attackers cannot erase it.