title: “The 3-2-1 Backup Strategy, Actually Implemented”
description: “Stop planning backups and start having them. A practical guide to building the 3-2-1 rule at home: local copy, NAS copy, off-site copy, automation and restore testing.”
date: 2026-09-20
draft: false
categories: [“Storage”]
tags: [“backup”, “3-2-1”, “NAS”, “cloud storage”, “data safety”, “rclone”]
summary: “What the 3-2-1 rule really means and how to implement it with free tools, cheap storage and a restore test you can actually pass.”
The 3-2-1 Backup Strategy, Actually Implemented
Everyone knows the rule: 3 copies of your data, on 2 different media types, with 1 copy off-site. Almost nobody has it. This guide is the implementation — which copies, which tools, which schedules, and how to verify it all works before you need it.
What the rule actually means
| Number | Meaning | Why |
|---|---|---|
| 3 copies | Your working copy + 2 backups | One copy is not a backup; two means a failure plus a restore gone wrong still leaves you safe |
| 2 media types | e.g. external drive + NAS/cloud | A dead laptop + a dead external drive from the same batch is rare — but a stolen bag takes out all local copies at once |
| 1 off-site | Cloud, or a drive at another house | Fire, theft and flood take out every copy in your home regardless of redundancy |
The off-site copy is the part most people skip, and it’s the part that actually protects you.
The three copies, concretely
Copy 1 — Your working data (the original)
This is your computer, phone, or the drive you edit photos from. Nothing to set up — but it’s not a backup. A backup is a separate copy you don’t touch every day.
Copy 2 — Local backup (different media)
The classic implementation is an external drive, because it’s genuinely a different medium from your internal SSD — and it’s cheap:
Windows: File History (built-in) or Veeam Agent Free for full image backups.
macOS: Time Machine — plug in a drive, it just works.
Linux:
rsyncor Timeshift for the system, and a simplersyncscript for data.
A full disk image (Veeam/Time Machine) is the one that saves you when a drive dies mid-project: boot the image, you’re back. Don’t rely on copying files only.
Copy 3 — The off-site copy (NAS → cloud, or a drive at work)
Two good implementations, pick based on your data size:
Option A — Cloud for the important stuff (photos, documents, passwords):
Backblaze B2 or Wasabi, ~$6/TB/month, billed by usage.
Use rclone to sync a folder nightly. Rclone is one command, free, and runs everywhere:
rclone sync /home/you/Photos remote:bucket/Photos --backup-dir remote:archive/Photos/\$(date +%F)
The --backup-dir line keeps old versions so an accidental delete is never instantly mirrored into the cloud.
Option B — A second drive off-site:
If your data is too big for cheap cloud (a 20 TB media library), keep a second external drive at a family member’s house or a work drawer, and swap it monthly. It’s manual, but a 20 TB cloud bill is its own disaster.
Automation: the part that makes it survive
Backups that require remembering are backups that stop. Set schedules that run without you:
| What | When | Tool |
|---|---|---|
| Phone photos | Nightly, on Wi-Fi | Google Photos / iCloud / Synology Photos |
| Computer files | Nightly at 2 AM | Veeam / Time Machine / rsync |
| NAS → cloud | Nightly at 3 AM | rclone cron job / Synology Hyper Backup |
| Critical folder, real-time | On change | Syncthing (self-hosted) or Dropbox |
And the single most useful setting: a failure alert. rclone, Veeam and most NAS apps can email you when a job fails. A backup job that silently failed for three months is exactly how people lose a year of photos.
The part nobody does: restore tests
A backup that can’t be restored isn’t a backup — it’s hope. Test on a schedule that matters:
Monthly: restore one random file or folder from the local backup and open it.
Quarterly: restore a full system image to a spare machine (or at least boot it in a VM).
Yearly: pull the off-site copy and confirm the cloud bucket isn’t empty, encrypted with a lost key, or syncing a corrupt directory.
Rclone makes the cloud check one line:
rclone lsl remote:bucket | tail -5 # confirm files are really there
Also write down your encryption keys and rclone config location. The #1 cause of “my backups are gone” after ransomware isn’t ransomware — it’s a lost password or a config file that was never backed up itself.
What it costs
| Setup | One-time | Monthly | Restore time |
|---|---|---|---|
| External drive + manual copy | ~$60–100 | — | Minutes–hours |
| External drive + NAS + free cloud tier | ~$300–600 | $0 | Minutes |
| External drive + NAS + paid cloud (1 TB) | ~$400–700 | ~$6–15 | Minutes |
| Two external drives, swapped monthly | ~$150–200 | — | Hours |
Start with the cheapest row today — an external drive and a schedule beats a perfect plan that hasn’t been started. Then grow into the NAS row when you’re ready; our home NAS guide covers choosing one, and the media server guide shows what else the box can do.
The starting point we recommend
A 2-bay NAS for your on-site copy, plus cloud for the important folders, is the 3-2-1 sweet spot for most households:
2-Bay NAS (Beginner-Friendly Brand)
The bottom line
Three copies, two media types, one off-site — implemented as: external drive tonight, NAS when you’re ready, cloud or a family member’s drawer for the third copy, nightly automation, and a monthly restore test. A backup you’ve never restored is a guess. A backup you’ve restored once is a system you can trust.
Not sure what needs backing up? Tell us what you store on the Contact page and we’ll suggest a tier that matches.