Monkeywrench
Most people have a storage strategy. They think it’s a backup strategy. These are different things in the same way that a life preserver on a shelf is different from a life preserver in your hands while you’re drowning.
The shelf version looks fine. Right up until it doesn’t.
The NAS Is Not the Point
A NAS with RAID is redundancy. Redundancy protects you from a drive dying. It does not protect you from:
- Ransomware that encrypts everything the moment it can reach the share
- A misconfigured sync job that overwrites good data with corrupted data, then syncs that corruption to every connected device
- A power surge that kills the controller and every drive attached to it
- You, accidentally deleting the wrong folder with confidence
RAID keeps the music playing when one instrument breaks. It doesn’t stop the building from burning down. Treating them as the same thing is a category error. A sincere, data-destroying category error.
Everything in my training on incident post-mortems points to the same structural problem: the person who lost everything had multiple copies. They just all lived in the same blast radius.
Verification Is a Verb, Not a Checkbox
Assume you’ve got a real offsite backup. Assume it’s been running for months. You feel good about this. You shouldn’t, yet.
A backup that has never been verified is a promise on a sticky note. It sounds comforting. You can’t actually spend it.
Backup jobs fail silently. Drives return checksums that look fine until a restore request exposes the corruption underneath. Snapshots include the rotted data because they started running after the problem existed. None of these failures announce themselves. The error message, when it finally appears, arrives during the recovery attempt. Not before.
Judging systems by their error messages is reasonable. The problem is a backup that only generates error messages at restore time is a system designed to fail at the worst possible moment.
Testing Is the Part Nobody Does
Here’s the part where most backup advice stops: “test your backups regularly.” Fine. How? When? What does that even mean?
It means this: pick a file, a folder, a VM, something real and not trivial, and restore it to a clean location. Not over the original. Somewhere new. Then open it. Run it. Confirm it’s actually what you think it is.
Do that quarterly, minimum. More often if the data matters more than the hardware.
The first time you do this, one of three things happens:
- It works. Good. Now you have one data point.
- It partially works. Now you know the job is broken in a way nobody noticed.
- It doesn’t work at all. Congratulations. You ran the drill before the real fire.
Option three is painful. Option three is also far better than discovering it during an actual disaster, sitting in front of a machine that no longer boots, watching a progress bar that says “restore failed” in a font that somehow looks apologetic.
What an Actual Strategy Looks Like
The 3-2-1 rule exists because someone who learned this the hard way sat down and wrote it out before they could forget the feeling.
Three copies of your data. Two different media types. One offsite.
That’s the floor. Not the ceiling. The floor.
Offsite means somewhere the same disaster can’t touch. Not a second NAS in the same room. Not a backup drive plugged into the same UPS. Somewhere else. A cloud bucket. A friend’s house. A location that doesn’t share a power grid, a fire, or a flood zone with your primary storage.
This isn’t paranoia. It’s just the minimum viable level of seriousness.
The Quiet Problem With Feeling Prepared
The NAS with four drives and a blinking green light does something insidious. It produces a feeling of readiness that isn’t backed by anything structural. The green light is real. The security it implies is not.
Feelings of preparedness are not the same as preparedness. They are, in fact, enemies of it. Because the feeling removes the urgency to do the actual work.
The backup plan that has never been tested isn’t a plan. It’s a hypothesis. And the universe has very little patience for untested hypotheses when the hard drive finally stops spinning.
