Every business has one file it cannot afford to lose — the accounts, the client list, the folder called Final FINAL v2. It usually lives in one place, and everyone quietly assumes it will be there tomorrow. Drives do not share that assumption. They die without warning, they leave in a stolen bag, and ransomware encrypts them by the thousand.
The 3-2-1 rule is the standard answer, and it fits in one sentence: keep three copies of your data, on two different kinds of storage, with one copy somewhere else. Everything else is detail — why each number exists, and the quiet ways it fails.
Why three, why two, why one
Each number exists because of a specific way businesses have lost everything.
Three copies means the working copy plus two backups. With a single backup, the day your main drive dies you are one corrupted file away from nothing — and backups do turn out to be corrupted, usually at the worst moment. The second backup is your margin.
Two kinds of storage exists because identical things fail in identical ways. Two matching drives bought the same day can die within weeks of each other, and the surge that kills the computer kills the backup drive plugged into it. Spread the copies across different media — internal drive, NAS, cloud — so no single failure reaches them all.
One copy offsite is the disaster clause. If the office floods or burns, the backup drive next to the server goes with it. It is also the ransomware clause: modern strains hunt for backups first, and a NAS on the same network is easy to find. The first 24 hours of a ransomware incident are often the discovery that the backups went too.
What it looks like at three sizes
One person, one laptop
The working copy is the laptop. Copy two is an external drive run by the tool already in the machine — File History on Windows, Time Machine on a Mac. Copy three is a cloud backup service working in the background. That is the whole rule for the price of a drive and a modest subscription; our backup walkthrough has the exact steps.
A five-person office
First get the files out of five separate laptops into one shared place — Microsoft 365, Google Workspace or a shared drive — so there is one thing to protect instead of five. Back that up to a NAS in the office, and have the NAS copy everything to cloud storage overnight. Server, NAS or cloud walks through which fits.
An office with a server
The server backs up nightly to a local backup device — fast at restoring whole machines — and that backup replicates to cloud storage. At this size the question stops being hardware and becomes time: how quickly a dead server must be running again decides what you buy.
Sync is not backup
OneDrive, Dropbox and Google Drive are mirrors, not backups. The thought experiment that settles it: delete a file on your laptop. Within a minute, sync has deleted it from the server and every other machine — the mirror is doing its job. When ransomware encrypts your files, sync uploads each scrambled version over the good one with the same diligence.
Sync services keep recycle bins and version history, which soften the blow, but they hold files for a limited window and were never built for restoring a whole company at once. A backup is different on purpose: a separate copy, on a separate system, that does not obey your mistakes.
Versioning and retention, in plain words
A backup that keeps only the latest copy of each file repeats the sync problem in slow motion — the damage overwrites the good version at the next run. Versioning means keeping older versions too, so you can reach back past the damage. Retention is how far back: a common shape is daily versions kept for a month and monthly versions for a year. Corruption and accidental deletions are often noticed weeks late, and a backup that only remembers yesterday cannot help you.
The copy ransomware cannot touch
Attackers know the backup is what stands between you and paying them, so deleting it is part of the playbook. The answer is a copy nothing on your network can alter. An offline copy — a drive plugged in for the backup, then unplugged and stored elsewhere — cannot be touched, because it is not connected to anything. Immutable cloud storage does the same job without the ritual: once written, the backup cannot be changed or deleted by anyone, including you, until a set period expires. Most business backup services offer it, and it is the setting most worth turning on.
Two questions that size the whole thing
Before buying anything, answer two questions in plain numbers. How long can the business be down — an hour, a day, a week? And how much work can you afford to lose — everything since last night, or nothing past the last fifteen minutes? The answers decide whether a nightly cloud backup is enough or you need a local device that can stand in for a dead server within the hour. Settle the two numbers and the shopping list writes itself.
Test the restore — quarterly, on the calendar
A backup you have never restored from is a rumor. The software saying "completed" tells you a job ran, not that the files come back or the right folders were in it. Put a fifteen-minute appointment in the calendar every quarter:
- Pick two or three files at random, ideally ones that changed recently.
- Restore them to a different location, not over the originals.
- Open them and check they are the current versions, not months old.
- Note how long it took — that is your first honest estimate of a real recovery.
Once a year, go bigger: restore an entire folder, or prove a server image actually boots. It is dull. It is also the only step that turns a backup from a hope into a fact.
The quiet ways backups fail
Data rarely dies for want of a backup plan. It dies because the plan stopped working and nobody noticed. The recurring offenders:
- The backup that ran full. The drive or storage quota filled up months ago, every run since has failed, and the warning emails went to a folder nobody reads.
- The offsite copy that travels with the laptop. A USB drive in the laptop bag is not offsite — it gets stolen in the same theft and drowned by the same coffee.
- The folder that was never added. The job faithfully protects the folders picked years ago; the new accounts share is not one of them.
- The expired credential. A password change quietly stops the cloud job, which errors politely into a log nobody opens.
- The one person who understood it. They set it up, they left, and the knowledge left with them.
Each one is caught in minutes by the quarterly restore test. Found the other way, it becomes a data-recovery bill — and recovery only works when the drive died politely.
Backups you never have to think about
Setting up 3-2-1 properly is a day's work for someone who does it often, and the checking afterward is exactly what managed monitoring is for. Koadi does both: technicians set up the drives, the cloud copy and the versioning end to end, then keep watching the jobs so a failure is fixed the week it happens. Post the problem free; vetted technicians pick it up, you set a fixed price or take bids, and payment sits in escrow until you approve the work. Remote setup covers every US state, with on-site help from local technicians when hardware needs plugging in.