Every File Backup and Sync Method Explained: 3-2-1, RAID, NAS and Cloud
Backup vs sync, the 3-2-1 rule, incremental and differential backups, RAID levels, NAS, cloud, tape, ransomware protection and a simple creator workflow.
On this page
Your drive dies. Or you delete the wrong folder. Or ransomware locks everything. Is your footage still safe? Many creators think they have a backup, when really they have a copy on the same desk, or a sync folder that deletes along with the original.
This guide explains every backup and sync method you’re likely to meet. Facts were checked in October 2026 against official sources where they exist: CISA’s #StopRansomware Guide, Apple, Microsoft, Google Drive, Dropbox and the Amazon S3 documentation. Numbers not from a named source are examples, and my advice marks a recommendation, not a rule.
Quick answer: what protects you from what
| I need to survive… | Use |
|---|---|
| A bad edit | Version history or snapshots |
| A dead drive | RAID, or better, a second copy |
| Deletion or ransomware | An independent, versioned, offline backup |
| Fire or theft | An offsite copy |
| The long haul | Verified archives, refreshed on a schedule |
Backup vs sync
Data disappears in ordinary ways: a drive fails, you delete or overwrite something, ransomware encrypts files, a laptop is stolen, or fire, flood or a power surge hits the room. A real plan covers all of them.
- A backup is a separate, independent copy you can restore after something goes wrong. If the copy can die, be deleted or be encrypted at the same moment as the original, it isn’t really a backup.
- Sync keeps two or more locations looking the same. Great for working across devices, but a deletion, a mistake or a damaged file is copied everywhere too.
- Version history keeps older states of a file so you can roll back an edit, but only within a window. Dropbox documents 30 days on some plans, 180 on others, and up to 365 on the highest.
- An archive is finished work moved to long-term storage. An archive that exists in only one place still needs its own backup.
Sync is not a backup. Trash folders and version history help, but they’re a safety net with an expiry date, not a copy you control.
Coming soon: sync vs backup, why Dropbox and Google Drive aren’t backups.
The rules
3-2-1
Keep 3 copies (the original plus two backups), on 2 different types of device or media, with 1 copy offsite. A 2012 paper written at Carnegie Mellon for US-CERT recommended it, and photographer Peter Krogh is often credited with popularising it (that credit comes from secondary sources).
Each number covers a different risk: several copies cover a single failure, different media cover a fault that hits one type (like a bad batch of drives), and the offsite copy covers fire, flood or theft.
Vendors have stretched it to 3-2-1-1-0:
| Part | Meaning |
|---|---|
| 3-2-1 | The classic rule |
| +1 | One copy offline or immutable |
| 0 | Zero errors, proven by test restores |
It’s a naming convention, not a formal standard, but the extra ideas match what CISA advises.
Coming soon: the 3-2-1 backup rule, with setups for every budget.
Full, incremental and differential
| Type | Copies | Run time | To restore you need |
|---|---|---|---|
| Full | Everything selected, every time | Slowest, biggest | Just that backup |
| Incremental | Changes since the last backup of any kind | Fastest, smallest | The last full plus every incremental after it |
| Differential | Changes since the last full | Grows until the next full | The last full plus the latest differential |
So incrementals are quicker to run, and differentials are quicker to restore. With incrementals, one damaged link in the chain can break later restores.
Snapshots and retention
A snapshot is a point-in-time view of a file system, so you can go back to how things looked an hour ago. Snapshots are quick and small, but one stored on the same disk dies with that disk. Apple documents hourly local snapshots, kept for 24 hours.
Retention is how long older backups are kept. Time Machine shows the idea:
| Backups kept | For |
|---|---|
| Hourly | The past 24 hours |
| Daily | The past month |
| Weekly | Older months (oldest deleted when the disk fills) |
Drives, RAID and NAS
Drive types
- Hard disk drive (HDD): spinning magnetic platters. Usually the most capacity for the money, so common for bulk backups. Moving parts can wear out or break if dropped.
- Solid-state drive (SSD): flash memory, no moving parts, faster and tougher. It can still fail, sometimes suddenly, and flash left unpowered for long periods can lose data, so don’t shelve one as your only copy.
- External drive (DAS): direct-attached storage over USB or Thunderbolt. Cheap and fast, but it sits next to your computer (so theft, fire or a surge can hit both) and you have to remember to plug it in.
RAID levels
RAID (redundant array of independent disks) combines several drives into one volume for speed, capacity or surviving a drive failure. Usable space below uses 8 TB drives as an example:
| Level | How | Min drives | Survives | Usable (example) |
|---|---|---|---|---|
| RAID 0 | Striping | 2 | None (raises risk) | 4 drives: 32 TB |
| RAID 1 | Mirroring | 2 | 1 drive | 2 drives: 8 TB |
| RAID 5 | Single parity | 3 | 1 drive | 4 drives: 24 TB |
| RAID 6 | Double parity | 4 | 2 drives | 4 drives: 16 TB |
| RAID 10 | Mirrors, striped | 4 | 1 per mirrored pair | 4 drives: 16 TB |
RAID 5 has no spare protection while it rebuilds, which is why RAID 6 is safer for big arrays, where a second drive can fail during a long rebuild.
RAID is not a backup, as Dell’s support page puts it. It protects against one thing: a drive failing. A deletion happens on every drive, ransomware encrypts the whole array, and a controller fault, fire or theft takes it all. RAID keeps you running. A backup lets you recover.
Coming soon: RAID 0, 1, 5, 6 and 10 explained for creators.
NAS
A NAS (network-attached storage) is a small always-on file server on your network. It holds one or more drives, often in RAID, and your computers see it as shared folders. It’s good for shared footage, project folders and backing up several machines.
- It shares files over SMB (Windows, and Macs speak it too) or NFS (popular on Linux and Unix). Speed depends on your network, so many creators use wired gigabit or faster Ethernet.
- Appliance or DIY? A ready-made box with empty drive bays is easier to set up and support. A build from a spare computer with open-source software is flexible, but you become the administrator.
- A NAS alone isn’t enough. It’s still one device in one room. Your working drive copies to the NAS, and the NAS copies offsite.
Cloud
The word “cloud” covers two different products: sync services keep folders matching, and backup services keep restorable history.
What the sync services’ own docs say:
| Service | Documented safety net |
|---|---|
| Google Drive | Trash is deleted forever after 30 days |
| Dropbox | Version history of 30 to 365 days, by plan |
| OneDrive | Restore the whole OneDrive to any point in the last 30 days (Microsoft 365 subscribers) |
Those are recovery windows, not a backup you control.
A cloud backup service runs in the background, uploads selected folders, keeps older versions and lets you restore later. Before you trust one, read two settings: how long deleted files and old versions are kept, and what gets excluded.
The first upload is the catch. As an example, 2 TB of footage over a 20 Mbps upload line takes about 9 days of nonstop uploading. Over a gigabit local network to a NAS, it’s about 5 hours. Plan your first copy locally.
Object storage
Object storage, like Amazon S3, stores data as objects in buckets, each with an ID and metadata. Many backup tools can use it as a target.
- Versioning: with it on, overwriting creates a new version, and deleting only adds a delete marker, so earlier versions can still be restored. Amazon notes you pay for every stored version, so set rules to expire old ones.
- Egress and retrieval costs: many providers charge separately for getting data back, and some archive tiers take hours. I won’t quote prices because they change. Check what a full restore would cost, not just the monthly fee.
Trusting one company: your files depend on one provider, one account and one password. Use a strong password and two-step sign-in, and don’t make the cloud your only copy.
Offline storage and tape
Cold storage means copies kept disconnected, in a drawer, a safe or another building. A disconnected drive can’t be reached by ransomware or deleted by a bad script, which is why CISA tells organisations to keep offline, encrypted backups. The trade-off: cold copies get stale unless you refresh them.
My advice for a solo creator: rotate two backup drives. One stays connected for regular backups, and the other lives offsite (at a friend’s place or in a safe-deposit box). Swap them on a schedule, and you get your offline copy and your offsite copy in one move.
Magnetic tape is the old archive medium that data centres still use. Cartridges sit on a shelf, unpowered and offline, so they’re air-gapped by design, but reading is sequential, so finding one file is slow. LTO (Linear Tape-Open) is the common standard. These capacities come from trade coverage, so check the LTO consortium for current specs:
| Generation | Native | Compressed (est.) |
|---|---|---|
| LTO-9 | 18 TB | 45 TB |
| LTO-10 (spec) | 36 TB | 90 TB |
| LTO-10 (first drives, one report) | 30 TB | 75 TB |
My opinion: tape is overkill for most solo creators. The drives are specialised and costly, and you need a compatible drive years later. It starts to make sense for studios with tens of terabytes to keep for years.
Built-in tools
Time Machine (Mac): Apple says it backs up apps, music, photos, email and documents to an external drive, copying only changed files after the first backup. Apple suggests a disk at least twice your Mac’s capacity, used only for backups, with an option to encrypt.
- The oldest backups are deleted when the disk fills, so a small disk shortens your history.
- If you encrypt, you need the backup password to restore. Lose it and the backup is useless.
- Eject the disk safely, or shut the Mac down first.
File History (Windows): backs up your libraries (Documents, Pictures, Videos, Music) to an external drive or network location. To recover, right-click a folder and choose Restore previous versions. Microsoft warns a restore replaces the current version and can’t be undone, so restore to a different location if you’re unsure.
What built-ins don’t cover: they usually back up to one drive on your desk. That’s not offsite, not offline, and a drive that stays plugged in is reachable by ransomware.
Keeping backups safe
- Bit rot (data degradation) is slow, silent corruption of stored data. A video may open with one glitchy frame, or not open at all, and a corrupted file can get backed up over the good copy.
- Checksums are short fingerprints of a file’s contents (SHA-256 is common). Change one bit and the fingerprint changes, so matching checksums prove a file is intact. File systems like ZFS and Btrfs can checksum automatically, and scrubs re-read data to catch errors early.
- Verifying checks a backup matches its source: compare checksums, or open a few random files, especially the biggest videos. It catches failed copies, but it doesn’t replace a restore test.
- Ransomware: CISA warns many variants look for accessible backups to delete or encrypt. A backup that’s always mounted, or a sync folder, can be encrypted with everything else.
- What CISA advises: keep offline, encrypted backups of critical data, regularly test their availability and integrity in a disaster-recovery scenario, and check for dropper malware before restoring so you don’t reinfect yourself.
- Immutable backups can’t be changed or deleted for a set period, even by an administrator. In S3 that’s Object Lock, a write-once-read-many model. In compliance mode, no one can delete a protected version early, not even the root user.
- Encryption protects a stolen drive or breached account, but lose the key and you may never restore. Keep the recovery key separately, on paper or in a password manager.
Test a restore. A backup you’ve never restored is a hope, not a backup. Restore a random folder to a different location and open the files, after setup and again on a schedule.
Watch for silent failures: an unplugged drive, a full disk, an expired password, or a job that has been failing for weeks. Turn on notifications, and check monthly when the last backup finished and whether it’s complete.
For creators
Size maths: file size is bitrate × time. As examples, a 100 Mbps camera file is about 45 GB per hour, and a 700 Mbps editing codec is about 315 GB per hour. Check your camera’s real bitrate in its specs or in MediaInfo.
Not everything needs the same protection (my advice):
| Tier | What | Protection |
|---|---|---|
| 1 | Raw footage and project files (can’t be recreated) | Full 3-2-1, versioned |
| 2 | Final exports and key assets (graphics, licensed music) | 2 copies, 1 offsite |
| 3 | Proxies, caches, renders (can be regenerated) | Skip, or keep local only |
Archiving a finished project: put the raw footage, project file and final export in one folder named by date and project. Verify it, store two copies (one offsite), and only then clear your working drive.
A simple solo setup (my advice):
| Step | Role |
|---|---|
| Fast work drive | Where you edit |
| Footage on a second drive | Same day, before erasing the card |
| Automatic local backup | To a NAS or external drive |
| Cloud backup service | The offsite copy |
| One rotating offline drive | The offline copy |
The routine:
| When | Do this |
|---|---|
| Each shoot | Copy footage twice before formatting cards |
| Daily | Let the automatic backup run |
| Monthly | Check the last backup date and test one restore |
| Yearly | Rotate the offsite drive, open old archives, replace ageing drives |
In one line: three copies, two kinds of media, one offsite, one offline or immutable, and zero surprises because you tested the restore.
Five mistakes to avoid
- Treating sync as a backup.
- Treating RAID as a backup.
- Keeping every copy in one room.
- Never testing a restore.
- Encrypting without keeping the key somewhere safe.
Try it: restore one file from your backup today, and open it.
