A storage area network (SAN) is a dedicated, high-speed network that gives servers block-level access to a shared pool of storage. Instead of relying only on their own internal disks, servers reach storage systems over the SAN and see the volumes presented to them much like locally attached drives.
That block-level view is the core of any SAN storage definition. Each server’s operating system formats and manages its volumes directly, which is what separates a SAN from file-sharing approaches. The fabric has one job: carrying block requests between servers and the systems that hold the data.
How a SAN works: components and data path
A SAN has three main building blocks, plus the management software used to configure them:
- Hosts: servers fitted with a host bus adapter (HBA) or a software initiator that sends storage commands over the network.
- Fabric: the switches and cabling that carry storage traffic between hosts and storage.
- Storage targets: arrays or storage servers that divide their capacity into volumes, identified by logical unit numbers (LUNs).
How far the fabric is separated from general user traffic is a design choice. Fibre Channel (FC) uses its own adapters and switches, so its traffic travels on a separate fabric. An iSCSI (Internet Small Computer Systems Interface) SAN runs over Ethernet, and its traffic can be isolated with dedicated switches or virtual LANs (VLANs), or share an existing network.
The data path follows those parts in order. An application writes data, the operating system issues a block-level request to what it treats as a local disk, and the adapter or initiator sends that request across the fabric. The storage target writes the blocks to the right volume and returns an acknowledgment along the same route.
Because capacity sits in shared storage systems rather than inside individual servers, it can be pooled. Administrators allocate volumes to the hosts that need them and add capacity at the storage system instead of inside each server, although whether that expansion is non-disruptive depends on the array. Pooling also makes shared volumes possible: several servers in a cluster can access the same storage when clustering software coordinates that access.
SAN protocols: Fibre Channel, iSCSI and NVMe over Fabrics
A SAN can run over several transports. The right one depends on the network you already have, your team’s skills and the workloads you need to support.
- Fibre Channel: a protocol built for storage networking, with dedicated adapters and switches.
- iSCSI: carries SCSI block commands over standard TCP/IP Ethernet networks, so it can use familiar Ethernet switching.
- NVMe over Fabrics (NVMe-oF): extends the NVMe (Non-Volatile Memory Express) command set across a network fabric, with transports that include Fibre Channel, remote direct memory access (RDMA) and TCP.
Is iSCSI SAN or network-attached storage (NAS)? It is SAN. iSCSI delivers block-level storage to a server, which then manages the file system itself. Running over Ethernet does not turn it into file sharing.
SAN vs. NAS vs. DAS
SAN, NAS and direct-attached storage (DAS) are different access models rather than better or worse versions of the same thing.
| Feature | SAN | NAS | DAS |
|---|---|---|---|
| Access level | Block | File | Block |
| Typical protocols | Fibre Channel, iSCSI, NVMe-oF | SMB, NFS | SATA, SAS, NVMe (local) |
| How servers see storage | As local disks (volumes) | As network shares or folders | As local disks |
| Sharing model | Shared pool allocated to many hosts | Files shared by many users and servers | Tied to one server |
| Example workloads | Databases, virtualization clusters, clustered applications | File shares, home directories, content collaboration | Single-server applications, local boot and scratch space |
| Skills needed | Storage networking, fabric and volume management | File server and permissions management | Server administration |
Which one fits depends on the workload. If an application needs raw volumes it can format and manage, such as a database engine or a hypervisor datastore, block access through a SAN matches that need. If people and applications mainly need to share files, NAS serves them directly through SMB (Server Message Block) or NFS (Network File System). One environment can use both, each for the workloads it suits.
What is storage area network technology used for?
The purpose of a storage area network is to give multiple servers shared, centrally managed block storage. That matters wherever several hosts must reach the same data, or where storage has to be managed independently of any single server. Example workloads include:
- Databases that need dedicated block volumes they control directly.
- Virtualization clusters, where hosts share volumes so virtual machines can move between them.
- Clustered applications that rely on shared disks for failover between nodes.
A concrete example of a SAN is a virtualization cluster whose hosts share volumes on a storage array over a Fibre Channel or iSCSI fabric. The value comes from shared block access, not from the network alone.
What running a SAN involves
A SAN adds operational work, and that work affects staffing and tooling.
Access control comes first. Zoning on the fabric limits which hosts can see which storage ports, and LUN masking on the storage system limits which hosts can see which volumes. A mistake in either can expose a volume to the wrong server, so changes need care and documentation.
Redundancy is a design choice rather than an automatic property. Multipathing gives a host more than one route to its storage, and duplicating components along that route removes single points of failure. These measures can reduce the chance that one failed part interrupts service, but parts still fail, and how quickly service recovers depends on the design and on the procedures your team has actually tested.
There is also ongoing maintenance: keeping firmware and drivers compatible along the whole path, monitoring paths and performance, and planning capacity before the pool runs short.
Conclusion
In practical terms, a storage area network turns shared storage into block volumes that your servers treat as their own. It fits workloads that depend on shared block access, provided your team can run a storage fabric with sound zoning, masking and multipathing. A sensible next step is to map your current workloads, existing network, team skills and expected growth, then weigh SAN, NAS and DAS against those needs.
FAQ
SAN storage gives multiple servers shared, centrally managed block volumes that they use like local disks. Databases and virtualization clusters are typical examples.
iSCSI is a SAN protocol, because it carries block-level storage commands over Ethernet rather than sharing files.
Isolation means keeping storage traffic apart from general user traffic. Fibre Channel provides it through its own fabric; with iSCSI it is a design choice, using dedicated switches or VLANs.
Neither is better in general. A SAN suits workloads that need block volumes, while NAS suits workloads that mainly share files.
