
SAN vs NAS: which storage fits your servers
SAN vs NAS is not a choice between good and bad but between two ways of presenting storage. A SAN gives your servers disks; a NAS gives your users folders. Once you see that difference, you usually also know which one you need, and whether you might need both.
A SAN provides block storage: the server sees a disk and puts its own file system on it. A NAS provides file storage: users and applications see shares over SMB or NFS.
If you run virtual machines across several hosts, you need shared block storage; if it is about folders and documents, a NAS will do. Many systems can now do both.
Neither is a backup. A copy on the same box, or on a box reachable with the same admin password, disappears with a failure or an attack.
SAN vs NAS: blocks versus files
The difference lies in what the storage offers the network. A SAN (storage area network) delivers blocks. The server is handed a chunk of storage that behaves like its own disk, formats it itself and decides what goes where. The storage system has no idea what a file is.
A NAS (network attached storage) delivers files. The device has its own file system and hands out shares. A user or application asks for a file, and the NAS works out where it lives and who may open it.
Almost everything else in this article follows from that. Block storage is what a hypervisor or a database wants: low latency, direct control, no layer in between. File storage is what people want: folders, permissions per group, and several users working in the same folder at once.
What runs over the network
A SAN usually speaks iSCSI over ordinary Ethernet, Fibre Channel over a dedicated network with its own switches, or increasingly NVMe over Fabrics. A NAS speaks SMB (Windows shares) or NFS (Linux and many hypervisors).
For both, storage traffic should not share a lane with office traffic. What we set up in practice:
- A separate network or VLAN for storage, so a large file copy in the office does not make the virtual machines stutter.
- Two paths. Two network ports per server, two switches and multipathing, so one cable or one switch cannot take the environment down.
- Enough bandwidth between server and storage. Gigabit is usually too tight for shared block storage under virtual machines; 10 or 25 Gbit/s is the common size there.
That network is the part most often missing from a storage quote. A fast storage system behind a slow or single link is a slow storage system.
When a SAN fits
A SAN is the logical choice as soon as several servers need to see the same disks. That is exactly what a virtualisation cluster asks for: if a host fails, another host restarts its virtual machines, and that only works if their disks sit on shared storage.
- Virtualisation on two or more hosts with failover or live migration.
- Databases and applications with many small writes, such as an ERP or finance system with many concurrent users.
- Environments where downtime costs money by the hour. A business SAN has two controllers, so a failed controller or a firmware update does not interrupt storage.
Be clear about what you get: a SAN does not hand out folders to users. If you also want a file server, you run it as a virtual machine on the SAN, or put a NAS next to it.
When a NAS fits
A NAS is strong where it is about files for people: project folders, drawings, scans, exports from an application. You get permissions per user or group, often tied to Active Directory or Entra ID, and previous versions of files through snapshots.
- A central file location for an office that does not (yet) keep everything in SharePoint or OneDrive.
- Large files you do not want to pull through the cloud every time, such as CAD, video or medical imaging.
- A backup target, provided that NAS is separate from what it protects. More on that below.
Where a NAS is weaker: carrying the disks of a virtualisation cluster under heavy load. It can be done, over NFS or iSCSI on the same box, but then one device decides for both your users and your servers.
What an SME with two or three servers usually chooses
For most SME environments the question is not “SAN or NAS” but how much shared storage you actually need. There are four routes, and each suits a different situation:
- One server with its own disks. No shared storage needed. Fine as long as you can live with the time a restore onto a second machine takes.
- Two or three hosts with a small SAN, over iSCSI or connected directly with SAS cables. The classic design for a failover cluster.
- Unified storage: one system that serves both blocks and files. Fewer boxes, but also one device everything depends on; choose a model with two controllers.
- Hyperconverged: the storage sits in the servers themselves and is shared by software. No separate storage system, but at least two or three identical nodes.
And more and more often the honest answer is that the files move to Microsoft 365 and storage is only needed for a few applications. The storage systems we supply and manage are on our page about business storage systems.
The mistakes we see most often
- The NAS as the only backup. A backup on a NAS in the same rack, reachable with the same domain password, goes down with ransomware. A backup belongs somewhere an attacker with your admin password cannot delete it; how to set that up is in our article on the 3-2-1 backup rule.
- Mistaking snapshots for a backup. A snapshot lives on the same disks. If the system fails, the snapshots go with it.
- A consumer NAS in the server room. One power supply, one controller, drives not built for continuous use, and firmware updates you have to chase yourself.
- Buying capacity and forgetting performance. Large, slow drives give lots of terabytes and few writes per second. Virtual machines notice first.
- No plan for end of support. A storage system without firmware updates or spare parts is a risk that looks like a working device in the rack.
What to ask before you choose
What a storage system costs is driven by a handful of things, and they belong in the question you put to a supplier:
- How much capacity, and how much growth over the years the system has to last.
- Which workload: virtual machines, a database, files, or a mix. That decides the type of drives, flash or spinning.
- Which availability: may an update or a fault interrupt storage briefly, or do you need a second controller?
- Which network comes with it: switches, cabling and server ports are not always in the quote.
- How long support and parts are available, and who keeps the firmware current.
A quote that answers these five can be compared. A quote with only a number of terabytes cannot.
Where our work stops
What we do. We help you choose, supply the storage system and the network around it, set it up, connect it to your servers and backup, and manage it afterwards: firmware, monitoring, capacity and timely replacement.
What we do not do. We are not the manufacturer: warranty, firmware bugs and replacement parts go through the manufacturer, even when we log the call for you. And physical data recovery from a damaged drive in a clean room is work for a specialist lab, not for us. That is why we would rather set up a backup that makes it unnecessary.
Questions we get about this
The five that come up most often.
What is the difference between SAN and NAS?
A SAN provides block storage: a server sees a disk and puts its own file system on it. A NAS provides file storage: users and applications see shared folders over SMB or NFS. A SAN is built for servers, a NAS for files.
Do I need a SAN for virtualisation?
Only if several hosts must be able to run the same virtual machines, for failover or live migration. With a single host, local disks will do. Hyperconverged storage, where the servers share their disks with each other, is an alternative to a separate SAN.
Can a NAS work as a SAN?
Many business NAS systems offer iSCSI alongside shares and then deliver block storage. That works for smaller environments. Bear in mind that one device then serves both your users and your servers, and preferably choose a model with two controllers.
Is a NAS a good backup?
A NAS can be a good backup target, but not if it is reachable with the same password as the systems it protects, or sits in the same rack. Keep at least one copy that an attacker cannot delete, and test the restore.
Do you recover data from a failed NAS?
We restore from backups and replace failed drives in an array that is still running. Physical data recovery from damaged drives in a clean room is not something we do; that is work for a specialist lab. The warranty on the device runs through the manufacturer.
Further reading
What this choice connects to.
Shall we match your storage to your servers?
Tell us how many servers you have, what runs on them and how much space you expect to need. We will say which storage fits and what network goes with it.
Practical IT knowledge in your inbox
New guides on management, security and the workplace, written by the people doing the work. No sales talk, and you can unsubscribe in one click.