Every time I onboard someone onto a storage team, I notice the same thing: the technology isn’t what trips people up first, it’s the vocabulary. Storage networking has its own dense dialect borrowed from SCSI, networking, and decades of vendor-specific marketing. I put this glossary together the way I wish someone had handed it to me on day one — organized by category, explained in plain language, with the “why it matters” context that most glossaries skip.
Why Terminology Matters So Much in Storage
Unlike general IT where a slightly wrong term still gets the point across, storage terminology carries precise technical meaning that affects design decisions. Confusing a “LUN” with a “volume,” or a “SAN” with a “NAS,” leads to real architectural mistakes. I’ve watched procurement teams order the wrong equipment because of exactly this kind of confusion, so I take vocabulary seriously.
Foundational Architecture Terms
SAN (Storage Area Network) — A dedicated network that provides block-level storage access, typically over Fibre Channel or iSCSI. Hosts see SAN storage as raw block devices (disks), not files.
NAS (Network-Attached Storage) — A storage system that provides file-level access over standard network protocols (NFS, SMB/CIFS). Unlike a SAN, the storage system itself manages the filesystem.
DAS (Direct-Attached Storage) — Storage physically connected to a single server with no network in between (internal disks, direct SAS-attached JBOD).
Unified Storage — A storage platform that serves both block (SAN) and file (NAS) protocols from the same underlying hardware, common in modern arrays from Dell EMC, NetApp, and HPE.
Software-Defined Storage (SDS) — Storage functionality (pooling, replication, tiering) implemented in software, decoupled from proprietary hardware — e.g., VMware vSAN, Ceph, Red Hat Storage.
Hyperconverged Infrastructure (HCI) — An architecture that combines compute, storage, and networking into a single software-defined platform running on standard x86 servers (e.g., Nutanix, VMware vSAN, Dell VxRail).
Block-Level Storage Terms
LUN (Logical Unit Number) — A logical, addressable unit of storage presented to a host by a storage array; from the host’s perspective, it looks like a raw disk.
Volume — A logical storage container that can map to a LUN, a filesystem, or a virtual disk depending on context.
RAID (Redundant Array of Independent Disks) — A method of combining multiple physical disks into a logical unit for redundancy and/or performance. Common levels I work with:
| RAID Level | Description | Fault Tolerance | Typical Use |
|---|---|---|---|
| RAID 0 | Striping, no redundancy | None | Performance-only, non-critical |
| RAID 1 | Mirroring | 1 disk failure | Boot volumes, small critical data |
| RAID 5 | Striping with single parity | 1 disk failure | General-purpose, cost-efficient |
| RAID 6 | Striping with double parity | 2 disk failures | Large-capacity drives, higher fault tolerance |
| RAID 10 | Mirrored + striped | Multiple, depends on layout | High-performance, high-availability |
Parity — Redundant information calculated from data blocks (via XOR) that allows reconstruction of lost data after a disk failure.
LUN Masking — Restricting which hosts can see a specific LUN, typically configured at the storage array level, complementing FC zoning.
Thin Provisioning — Allocating storage capacity on-demand rather than reserving the full logical size upfront, improving utilization efficiency.
Thick Provisioning — Allocating the entire requested capacity upfront, guaranteeing space availability but reducing flexibility.
File-Level and Protocol Terms
NFS (Network File System) — A distributed file system protocol, primarily used in Unix/Linux environments, that lets a client mount remote storage as if it were local.
SMB/CIFS (Server Message Block / Common Internet File System) — The file-sharing protocol standard in Windows environments; modern versions (SMB3) include encryption and multichannel support.
iSCSI (Internet Small Computer Systems Interface) — A protocol that encapsulates SCSI commands within TCP/IP, allowing block storage access over standard Ethernet networks.
FCP (Fibre Channel Protocol) — The protocol mapping layer that carries SCSI commands over a Fibre Channel fabric.
NVMe-oF (NVMe over Fabrics) — A protocol extending NVMe’s low-latency, high-queue-depth command set across a network fabric (Fibre Channel, RoCE, TCP), rather than being limited to a local PCIe bus.
Fabric and Network Terms
WWN / WWPN / WWNN — World Wide Name identifiers used in Fibre Channel; WWNN identifies the device, WWPN identifies a specific port (covered in depth in my Fibre Channel article).
Zoning — Restricting which FC devices can communicate with each other within a fabric, similar in purpose to a network ACL.
VSAN (Virtual SAN — Cisco context) — A logical partition of a physical Fibre Channel fabric into isolated virtual fabrics (not to be confused with VMware vSAN, an entirely different HCI product that unfortunately shares the acronym).
MPIO (Multipath I/O) — Software that manages multiple physical paths between a host and storage, providing failover and load balancing.
ALUA (Asymmetric Logical Unit Access) — A SCSI standard that lets a storage array tell a host which paths to a LUN are “optimized” (owning controller) versus “non-optimized,” so multipathing software can route I/O efficiently.
HBA (Host Bus Adapter) — A physical adapter card in a server that connects it to a SAN (FC HBA) or, more loosely, to storage in general (SAS HBA).
CNA (Converged Network Adapter) — An adapter capable of carrying both standard Ethernet and FCoE traffic on the same physical port.
Performance and Capacity Terms
IOPS (Input/Output Operations Per Second) — A measure of how many read/write operations a storage system can process per second; critical for transactional workloads like databases.
Throughput — The amount of data moved per unit of time (MB/s or GB/s), more relevant for sequential, large-block workloads like backups or video streaming.
Latency — The time delay between issuing an I/O request and receiving a response, usually measured in milliseconds or microseconds; the metric that matters most for user-perceived application performance.
Queue Depth — The number of outstanding I/O requests a device or path can handle simultaneously before queuing additional requests.
Deduplication — Eliminating redundant copies of data at the block or file level to save capacity, common in backup targets and modern all-flash arrays.
Compression — Reducing the size of stored data using algorithms (LZ4, zstd, etc.), often paired with deduplication in modern storage platforms.
Tiering — Automatically moving data between storage tiers (e.g., NVMe flash, SAS, cold archive) based on access frequency, to balance performance and cost.
Data Protection and Availability Terms
Snapshot — A point-in-time, space-efficient copy of a volume or LUN, typically using redirect-on-write or copy-on-write techniques.
Replication — Copying data from one storage system to another, either synchronously (real-time, zero data loss, distance-limited) or asynchronously (near-real-time, longer distance, small potential data loss window).
RPO (Recovery Point Objective) — The maximum acceptable amount of data loss, measured in time, in a disaster scenario.
RTO (Recovery Time Objective) — The maximum acceptable time to restore service after a disruption.
Erasure Coding — A data protection method (common in object storage and scale-out systems) that splits data into fragments with added redundancy, tolerating multiple simultaneous failures with better capacity efficiency than traditional RAID at scale.
Failover / Failback — The process of switching operations to a redundant system after a failure (failover), and returning to the original system once it’s restored (failback).
Management and Protocol Support Terms
Fabric Login (FLOGI) — The process by which an FC device registers itself with the fabric and receives an FCID.
Discovery Domain — In iSCSI, a logical grouping (via iSNS) that controls which initiators and targets can discover each other.
iSNS (Internet Storage Name Service) — A discovery service for iSCSI environments, functioning similarly to the FC Name Server.
CHAP (Challenge Handshake Authentication Protocol) — An authentication mechanism used in iSCSI to verify initiator/target identity before allowing a session.
QoS (Quality of Service) — Policies that prioritize or limit bandwidth/IOPS for specific traffic classes or workloads to ensure predictable performance.
Object and Cloud Storage Terms
Object Storage — A storage architecture that manages data as discrete objects (data + metadata + unique identifier) rather than files or blocks, accessed via APIs like S3; used heavily for unstructured data at scale.
Bucket — A logical container for objects in an object storage system (S3 terminology).
Cloud Tiering — Automatically moving cold or infrequently accessed data from on-premises storage to a cloud object storage tier to reduce local capacity costs.
Gateway Appliance — A device or virtual appliance that translates between on-premises protocols (NFS/SMB) and cloud object storage APIs, enabling hybrid cloud storage architectures.
A Practical Glossary Table for Quick Reference
| Term | One-Line Definition |
|---|---|
| SAN | Block storage network |
| NAS | File storage over the network |
| LUN | A logical unit of block storage |
| RAID | Disk redundancy/performance grouping |
| WWPN | Unique FC port identifier |
| Zoning | FC device-to-device access control |
| MPIO | Multiple path management for storage |
| IOPS | I/O operations per second |
| RPO | Max tolerable data loss window |
| RTO | Max tolerable downtime |
| Snapshot | Point-in-time data copy |
| Deduplication | Removing redundant stored data |
| Erasure Coding | Distributed redundancy for scale-out storage |
Frequently Asked Questions
What’s the real difference between a LUN and a volume? A LUN is specifically a block storage construct presented over SAN protocols; a “volume” is a broader term that can refer to a LUN, a filesystem mount point, or a virtual disk depending on the platform.
Is NAS “slower” than SAN? Not inherently — modern all-flash NAS platforms can outperform older SAN arrays. The real difference is access method (file vs. block), not a guaranteed performance gap.
What’s the difference between synchronous and asynchronous replication? Synchronous replication confirms the write on both the source and target before acknowledging it to the host (zero data loss, but distance-limited by latency); asynchronous replication acknowledges the write locally first, then replicates shortly after (supports longer distances, small potential data loss).
Why do vendors use different names for similar concepts? Largely marketing and historical reasons — for example, “VSAN” means something completely different at Cisco (FC fabric virtualization) versus VMware (an HCI storage product), and I always double-check which one is meant in context.
Summary
Storage networking vocabulary spans block vs. file architectures, fabric and protocol terms, performance metrics, and data protection concepts — and precision in these terms genuinely affects design and troubleshooting outcomes. Building fluency in this vocabulary is, in my experience, the single fastest way to level up as a storage administrator, because nearly every vendor whitepaper, support case, and architecture diagram assumes you already know it.
References
- SNIA Dictionary of Storage Networking Terminology (snia.org)
- Dell EMC storage terminology and glossary resources (dell.com)
- NetApp glossary and documentation center (netapp.com)
- VMware storage and vSAN terminology guides (vmware.com)
- IBM Storage terminology reference (ibm.com)