Operate//guide // IT management
NAS or server for a small business? RAID is not backup
What a NAS does and what a server does, when a NAS is enough, what RAID 1 and RAID 5 mean and why mirrored disks do not replace a backup.
Updated: 17.11.2026 · ~7 min read
People who search for “nas vs server” or compare Synology models usually have the same problem: the company’s files are scattered across a few computers. For many companies of 5–20 people, a NAS is enough. For others, it would be a saving that gets paid for later. The difference lies in what has to run on the device, not in how much data you have.
What a NAS does
NAS stands for Network Attached Storage. It is a small computer with an operating system built for one job: holding disks and making them available on the network. Staff see shared folders on it, and the administrator decides who can access what.
Besides file sharing, a current NAS can receive scanned documents from the printer, keep earlier versions of files through snapshots, sync a folder with a cloud service and act as the destination for backups from the computers. Many models also run additional apps or containers. The limit shows when you need software that requires Windows Server, or more computing power than an office NAS processor has.
What a server does
A server is a general-purpose computer with a server operating system (Windows Server or a Linux distribution) that runs services for the rest of the company. Shared files are only one of them. On a small company’s server you will usually find:
- the database of the accounting, inventory or ERP software, in its client-server edition;
- centralised user accounts (an Active Directory domain), with policies applied to every computer;
- Remote Desktop sessions, through which staff work from home in the office applications;
- virtual machines, when several separate systems have to share the same hardware.
The hardware differs too. A business server usually has error-correcting (ECC) memory, disks that can be swapped without shutting the machine down, sometimes two power supplies, and a remote management board that lets you restart or reinstall it without a trip to the office.
Then there is licensing. Windows Server is licensed per processor core, plus client access licences for users or devices (CALs) and, for Remote Desktop, separate RDS access licences. What a correctly licensed server means and how to check what runs on yours is explained in the guide on unlicensed server software.
When a NAS is enough
A NAS is usually enough if the company matches almost all of the situations below:
- the shared data is mostly documents: quotes, contracts, drawings, photos, scans;
- the accounting or inventory software runs in the cloud, on a single computer or at the external accountant’s office;
- email is already with a provider (Microsoft 365, Google Workspace or another), not on an in-house server;
- you do not need centralised accounts with policies enforced on every computer;
- people working from home only need files, not applications running in the office.
A scenario: a design office with 8 people keeps its drawings on a colleague’s computer, shared on the network. When that computer is off or freezes, nobody can work. A NAS with two disks in RAID 1, a separate account for each employee and daily snapshots solves the problem without server licences.
When you need a server
A server becomes necessary when software requires a service that runs continuously, on the operating system the vendor specifies. Most often this is accounting or inventory software used over the network: the database runs on one computer and the other workstations connect to it. Many vendors require Windows for this and do not support placing the database directly in a network folder, because two workstations writing to the same file over the network can corrupt it. Check the requirements in the software documentation before choosing the hardware.
Another scenario: a distribution company with 15 employees runs its inventory software in client-server mode, has three sales agents who issue invoices from the field over Remote Desktop, and wants passwords to be changed from a single place. A NAS is not enough here. The right fit is a server for applications and accounts, and the NAS, if there already is one, becomes the backup destination.
RAID 1 and RAID 5 in brief
RAID means several disks combined so that the device keeps working if one of them fails. For a small company, two variants matter.
RAID 1: mirroring
Two disks with exactly the same data. If one fails, the other takes over, and after the replacement the data is copied to the new disk. Usable capacity is that of a single disk: two 4 TB disks give you 4 TB. It is the usual choice for a two-bay NAS.
RAID 5: parity
At least three disks. Besides the data, the device writes parity information, spread across all disks, from which it can rebuild the contents of any lost disk. Usable capacity is the sum of the disks minus one: four 4 TB disks give 12 TB. It still survives only one failed disk, and writes are somewhat slower because of the parity calculation.
The risky moment: the rebuild
Once you replace the failed disk, the device reads all the other disks to rebuild it. On large disks this can take from a few hours to a few days, during which RAID 1 and RAID 5 have no redundancy left. If another disk fails or read errors appear, the rebuild can fail, and with RAID 5 you can lose the whole volume. That is why RAID 6, which survives two failed disks, or RAID 10 is preferred for large or numerous disks.
Whatever the variant, a few things are worth doing: disks designed for continuous operation in a NAS or server, email notifications when a disk reports errors, and a UPS that shuts the device down cleanly during a power cut.
Why RAID is not backup
RAID protects you against one thing only: a disk failure. Everything that happens at file level is written immediately to all disks. A folder deleted by mistake disappears from both disks of a RAID 1. Ransomware that encrypts the shares encrypts the data on all disks at once. And if the NAS is stolen, catches fire or its motherboard dies, it does not help that the disks were mirrored.
Snapshots do not replace a backup either. They are very useful when someone overwrites a file, but they sit on the same device and can usually be deleted by anyone who gains administrator access to it. A backup is a separate copy, in another place and with different credentials. How to build one that survives ransomware is covered in the guide Business backup: the 3-2-1 rule, and how to check that you can actually restore from it, in the article on the restore test.
Before you buy
The answers to the questions below matter more than the spec sheet:
- which programs have to run on the network and which operating system their vendors require;
- how much data you have today and how much it grows in a year, so the disks have room for a few years;
- who works outside the office and what they need to open from there;
- where the backup copy will live, separate from the NAS or server;
- who looks after the device: updates, disk alerts, the accounts of employees who leave.
Choosing, installing and maintaining a NAS or a server is usually part of IT management.
FAQ
NAS, server and RAID
Can I put the accounting software on a NAS?
It depends on the software. If the database needs a service that runs on Windows, a NAS is not built for that, and a database placed only in a shared folder and opened by several workstations risks getting corrupted. Read the network installation requirements in the vendor’s documentation.
Which RAID should I choose for a two-disk NAS?
RAID 1. You get the capacity of one disk, and the device keeps working if one of them fails. With four or more disks, RAID 5, RAID 6 or RAID 10 come into play, depending on capacity and on how long you can afford to work without redundancy during a rebuild.
If I have RAID, do I still need a backup?
Yes. RAID does not protect you from deletions, ransomware, theft or fire, because every change is written to all disks at once. You need at least one separate copy, in another place and with different credentials.
Can an ordinary PC be used as a server?
For a few users and software that supports it, it can work. It has limits, though: Windows 10 and 11 accept at most 20 simultaneous connections to shared folders, their licence does not allow several Remote Desktop sessions at the same time, and the hardware lacks the redundant components of a server.
How long does a RAID rebuild take after replacing a disk?
From a few hours to a few days, depending on disk capacity, the RAID type and how heavily the device is used in the meantime. Data stays accessible, but slower. With RAID 1 and RAID 5, the array cannot survive another disk failure until the rebuild finishes.
Not sure a NAS is enough?
Tell us how many people work on shared data and which programs you use on the network, and we can discuss which option makes sense for your company.