What a backup is
A backup is a separate copy of data, kept so that you can recover if the original is lost, damaged or changed by mistake. The key word is separate. A copy that sits on the same disk, in the same account or in the same building as the original often disappears in the same incident.
People lose data in ordinary ways: a dropped laptop, a failed drive, an accidental deletion, a bad software update, a stolen device, or malicious software such as ransomware attacks that encrypt files and demand payment. Each of these calls for a copy that was out of harm’s way.
Backup, sync and archive are different things
These terms are often mixed up.
- Sync services keep folders matching across devices. If you delete a file in one place, the deletion usually spreads everywhere. That is convenient, but it is not by itself a backup.
- Backups keep earlier states, so you can go back to how things looked yesterday or last month.
- Archives move data you rarely need into long-term storage, often to free up space. An archive may be the only copy, which is a different risk.
Some cloud services combine features, including file version history, so check what yours actually keeps and for how long.
Types of backup
Backup software generally uses three approaches.
Full backup. Everything is copied each time. It is simple and fast to restore, but takes the most time and space.
Incremental backup. After one full copy, each later run saves only what changed since the previous backup of any kind. It is small and quick, but a restore may need the full backup plus every increment in order.
Differential backup. Each run saves everything changed since the last full backup. It grows over time, though a restore needs only the full copy and the latest differential.
A hypothetical example with illustrative sizes: suppose a folder holds 100 GB, and about 2 GB changes each day. A weekly full backup copies 100 GB. Daily incrementals copy roughly 2 GB each. Daily differentials would copy about 2 GB on the first day, about 4 GB on the second, and so on, since each includes all changes since the full copy. These figures are made up to show the pattern, not typical values.
Schedules, retention and recovery goals
Two ideas help set a schedule.
- Recovery point objective (RPO): how much recent work you can afford to lose. If backups run nightly, you might lose up to a day of changes.
- Recovery time objective (RTO): how long you can afford to be without the data while restoring.
A personal photo library may be fine with weekly backups. A business processing orders all day usually needs far more frequent copies. Databases are often protected with a mix of periodic full copies and continuous logs of changes, which allow recovery to a particular moment. How data is organized in the first place, covered in the introduction to database storage, affects how this is done.
Retention is how long old backups are kept. Keeping only the latest copy means a problem you notice late, such as slowly spreading corruption, may already be baked into every copy. Keeping several generations gives you somewhere clean to return to, at the cost of storage.
The 3-2-1 rule of thumb
A widely used guideline is 3-2-1: keep three copies of your data (the original plus two backups), on two different kinds of storage, with one copy offsite. Offsite might mean a cloud service or a drive stored at another location. The rule is a starting point and not a guarantee, but it captures the main idea of avoiding a single point of failure. Some people add an offline or write-protected copy so malware cannot reach it.
Why backups fail
Backups rarely fail loudly. More often, they quietly stop working long before anyone needs them.
- Never tested. The most common story is discovering at restore time that the files are incomplete, unreadable or in a format nobody can open.
- Silent job failures. A schedule stops because a drive filled up, a password changed or a device was unplugged, and no one noticed.
- Missing data. Important folders or databases were never included in the backup selection.
- Same-location copies. Fire, theft, a power surge or a single account compromise destroys both original and backup.
- Connected backups. If the backup drive or share is always reachable, ransomware can encrypt it along with everything else.
- Corrupt or already-damaged data. A backup faithfully copies whatever it is given, including errors.
- Lost keys. Encrypted backups cannot be opened without the password or key, and losing it can make the data permanently unreadable.
- Too slow to restore. Copying many terabytes back over a modest connection can take far longer than expected.
- Aging media and software. Old drives wear out, and old backup formats may not be readable by newer systems.
Testing is what makes a backup real
A backup is only proven when you have restored from it. Good habits include:
- Restore a few files regularly and open them to confirm they are intact.
- Occasionally rehearse a full recovery on a spare machine, timing how long it takes.
- Turn on alerts for failed or missed jobs and check logs.
- Write down the restore steps, so they are available when you are stressed and the usual systems are down.
- Review what is included whenever you add new folders, apps or devices.
Cloud and backups
Storing data with a provider of cloud infrastructure does not automatically mean it is backed up the way you want. Providers typically protect their own hardware, but your own mistakes, such as deleting a folder, may be replicated everywhere. Many services separate durability (data is unlikely to be lost from hardware failure) from backup (you can recover an earlier version). Read the terms and settings to see which you have.
A simple starting point
For an individual, a reasonable plan is an automatic backup to an external drive, plus an encrypted copy in an offsite service, plus a restore test every few months. The details will depend on how much your data matters and how much effort you can give. What counts is that the copies are separate, current, and known to work.