Common Myths About RAID and Backups

A friend called me at 11pm on a Sunday because his Synology DS920+ had a red light on one drive. He was calm, almost cheerful, because he had four disks in a RAID array and figured the whole point was that a dead drive was no big deal. Two days later, during the rebuild, a second drive dropped out and the volume went read-only, then unmountable.

He lost about six years of family photos.

The painful part is that none of it needed to happen. He had done nothing lazy or stupid. He had simply believed a handful of things about RAID and backups that sound completely reasonable and are completely wrong.

So let me walk through the myths I hear most often, the ones that cost people real data, and give you the correction for each. Some of these I believed myself back when I built my first box in a spare tower case.

Myth 1: RAID is a backup

This is the big one, the mother of all storage misconceptions, and it kills more home data than any hardware fault ever will. RAID keeps your NAS available when a disk dies. It does not keep your files safe.

Those are two different jobs. Availability means the box stays online and you keep working while you swap a bad drive. Safety means that when something deletes, corrupts, or encrypts your files, you have a second untouched copy somewhere else.

RAID gives you the first and nothing of the second.

Think about what RAID actually protects against: one (or sometimes two) drives failing mechanically. That is a narrow slice of how people lose data. It does absolutely nothing about the far more common causes.

Here is the honest list of things RAID will not save you from: accidental deletion, a botched folder sync that overwrites good files with empty ones, ransomware, a corrupted database, a power surge that fries the whole unit, theft, fire, flood, or you fat-fingering rm -rf in the wrong directory. I have personally watched three of those happen to people who swore they were covered because they had RAID.

The industry has a tidy rule for this called the 3-2-1 backup strategy: three copies of your data, on two different types of media, with one copy off-site. Your RAID array is copy number one. It is not the backup. It is the thing the backup protects.

Myth 2: A second drive in the array counts as my backup copy

People point at a mirror (RAID 1) or a parity array and say the redundant disk is their backup. It feels true. There is literally a second physical drive holding a copy of the bits.

The problem is that the second drive is not independent. It updates in real time with the first. A backup, by definition, is a copy that does not change when the original does, so you can roll back to a known-good moment.

Your mirror has no yesterday. It only has now, and now includes whatever mistake you just made.

What a real backup gives you that RAID cannot

Look at that column of noes. That is the exact set of disasters that actually erase home data, and RAID answers none of them.

Myth 3: RAID 5 with big drives is perfectly safe

This one is more technical, and it bites people who did their homework but read old advice. The concern is the rebuild.

When a drive fails in RAID 5, replacing it forces the array to read every single sector on all the remaining drives to reconstruct the missing data. With modern 12TB, 16TB, or 20TB disks, that is many hours (sometimes over a day) of the surviving drives running flat out, at exactly the moment they are already stressed and often the same age and model as the one that just died.

I am not telling you RAID 5 is forbidden. On a four-bay home NAS with 4TB drives it is often fine. But if you are running large disks, consider RAID 6 or Synology SHR-2, which tolerate two failures and survive a read error during rebuild. And whatever you pick, it still is not a backup.

Myth 4: Cloud sync like Dropbox or Google Drive is my backup

Sync is not backup. I will say it again because it is genuinely counterintuitive.

Sync services mirror your current state to the cloud. If you delete a file locally, the sync dutifully deletes it in the cloud too. If ransomware encrypts your folder, the encrypted versions sync up and overwrite the good ones.

The whole design goal of sync is to make everything match, which is the opposite of what a backup needs to do.

Now, most cloud services keep deleted files and old versions for 30 days, which has saved plenty of people. But 30 days is not much runway if you do not notice a problem quickly, and free tiers often keep far less. A real backup tool keeps versions on your schedule, not theirs.

Myth 5: My NAS does snapshots, so I am fully covered

Snapshots are excellent and I use them daily. On a Btrfs volume, a snapshot lets you pull back a file exactly as it was hours or days ago, which handles deletion and most ransomware beautifully.

But a snapshot lives on the same NAS, on the same disks, in the same building.

If the unit is stolen, catches fire, gets hit by lightning, or its volume corrupts at the filesystem level, your snapshots vanish along with everything else. They are a fantastic first line of defense and a terrible last one. Pair them with an off-site copy and you finally have something real.

Myth 6: Backups I never test still work

An untested backup is a hope, not a plan. I have opened backup folders that were faithfully copying an empty source directory for months because a mapped drive had silently disconnected.

The job ran green every night. It backed up nothing.

Getting yourself genuinely covered

None of this requires an enterprise budget. Here is the setup I recommend to nearly every home NAS owner, and roughly what it costs.

• Keep your RAID for uptime, but stop thinking of it as protection. It is convenience.
• Add an off-site copy. A cloud backup like Backblaze B2 or a rotated external drive kept at a relative's house covers fire and theft. Cloud storage for a few hundred gigabytes runs a handful of dollars a month.
• Turn on versioned snapshots if your NAS supports Btrfs. Free, and it saves you from yourself weekly.
• Test a restore monthly. Free, and non-negotiable.

That is three copies, two media types, one off-site. The rule that would have saved six years of my friend's photos, in about an hour of setup.

The reframe that keeps your files alive

Stop asking whether your data is redundant and start asking whether it is recoverable. Redundancy is about the box staying up. Recovery is about you getting your files back after something goes wrong, and something always eventually goes wrong.

RAID is a seatbelt for one specific kind of crash. A backup is the ambulance, the hospital, and the spare car for everything else.

Related reading: NAS Hard Drive Failure: Signs, Causes and Fixes, Synology SMART Warning: What It Means and Next Steps and How to Replace a Failed Drive in Synology DSM.

Build both, test the backup, and that Sunday-night red light becomes a shrug instead of a funeral.