Technical Anshu
Tutorials Blog
🛡️ Data Safety

RAID Is Not a Backup Snapshot and External Copy Rule for File Servers

RAID answers one question: what happens when a disk dies? Your data faces at least four other questions. This page separates the protections so each threat gets the right answer.

✍️ Technical Anshu💾 RAID / ZFS / Backup⏱️ 6 minute read
🎥 Full video — all 3 file server methods built and compared, step by step

Every file server conversation eventually produces the sentence “we have RAID, so we have backup”. It feels logical — two disks, two copies. But a mirror copies everything instantly, including the disaster: delete a folder and it vanishes from both disks in the same second; ransomware encrypts both copies with the same enthusiasm. RAID is about uptime. Backup is about history. A safe server needs both, plus one copy the server itself cannot reach.

Which threat does each protection actually stop
One disk diesMirror / RAID COVERED
File deleted or overwrittenSnapshots COVERED
Ransomware encrypts sharesSnapshots + offline copy IF KEPT OFFLINE
Server stolen, fire, floodExternal / offsite copy only RAID USELESS
✔ Uptime, undo and disaster recovery are three different jobs
✔ A protection you have never test-restored is a hope, not a backup
The protection map every file server in this series is built around.
1

What RAID and Mirrors Actually Buy You

A mirror (Windows Storage Spaces, TrueNAS mirror, RAID 1) writes every file to two disks at once. When one disk fails — and disks fail on their own schedule, not yours — the server keeps running and you swap the dead disk at a calm moment. That is genuinely valuable: no downtime, no emergency. RAIDZ on TrueNAS does the same trick across several disks. Buy it for continuity. Just never let the word "redundancy" get translated into the word "backup" in your head — they solve different problems.

2

Snapshots — the Undo Button for Human Mistakes

Most data loss in a small office is not hardware at all; it is a person deleting the wrong folder, saving over the master spreadsheet, or malware encrypting what it can reach. Snapshots answer all three by keeping frozen versions — hourly for a day, daily for a month is a typical office pattern. On TrueNAS they are nearly free in space. On Windows, scheduled Volume Shadow Copies give shared folders the same "Previous Versions" tab. Either way, recovery is a right-click, not a restore project.

3

The Limit of Snapshots — Same Box, Same Fate

Snapshots live on the same disks, in the same machine as the data. A dead pool, a stolen server, a fire — snapshots die with everything else. Worse, ransomware that reaches the server with admin-level access can delete snapshots deliberately; that is why the third layer is non-negotiable: at least one copy that is not continuously connected to the server or the network — a rotated external drive kept offline, or a scheduled copy to another location. Malware cannot encrypt a disk that is sitting unplugged in a drawer.

4

The Rule to Build By — 3, 2, 1

The classic rule survives because it works: 3 copies of anything important, on 2 different kinds of storage, with 1 copy offsite or offline. On a small-office budget it looks like this: the live data on the server (copy 1), snapshots plus a nightly copy to a second disk or NAS (copy 2), and a weekly external drive that leaves the building or stays unplugged (copy 3). You do not need enterprise software for any of this — TrueNAS replication, Windows backup tools or even a scheduled robocopy to the external drive all count, as long as they actually run.

5

Test the Restore, or It Does Not Exist

Once a quarter, pick one real folder and restore it to a test location from each layer: yesterday's snapshot, last night's copy, last week's external drive. Time it. This fifteen-minute ritual is the difference between offices that shrug at disk failures and offices that discover, mid-crisis, that the backup job failed silently in March. Automate alerts if your platform offers them, and treat any failed backup job as a same-day fix, not a footnote.

🔴 The one-sentence version: RAID keeps you working today, snapshots give you yesterday back, and only an offline or offsite copy survives the worst day. A file server missing any of the three is not finished — it is waiting.

🎥 Watch the Full Video

The video's TrueNAS method shows snapshots being created and a file being rolled back live — the clearest way to understand what this page describes.

▶️ Watch Full Video
❓ Common Questions (FAQ)

Is a two-disk mirror enough for a very small office?

It is the right start and the wrong finish. Add snapshots the same week (they cost almost nothing) and one external drive rotated weekly. That combination — mirror plus snapshots plus offline copy — covers disk death, deletion, ransomware and theft for the price of a single extra drive.

How often should the external copy run?

Ask how much work the office can afford to redo. A nightly copy means a disaster costs at most one day of work; a weekly copy risks five days. Most small offices land on nightly-to-a-second-device plus weekly-offline. Whatever you choose, write the schedule down and check it monthly.

Can cloud storage be the offsite copy?

Yes — syncing an encrypted copy to cloud storage is a valid offsite layer and removes the discipline of rotating drives. Two cautions: a plain two-way sync also syncs deletions and ransomware encryption, so use a backup tool or versioned cloud backup rather than a mirror sync, and confirm the restore speed is acceptable for your data size.

This guide is based on the Technical Anshu file server video • Stuck on any step? Ask in the video comments — you will get a reply there 🙏