Cloud backup or a drive on the shelf? Which one survives which disaster

The two options fail in completely different ways, which is the actual argument for having both. What matters is which failure you are protecting against.

Article details
AuthorCorinne Adeyemi
SectionInnovation
Published
Length868 words · 4 min
A portable external hard drive and a coiled cable resting on a desk beside a closed notebook
Fig. 1 — A portable external hard drive and a coiled cable resting on a desk beside a closed notebook

The choice is usually framed as cloud versus local, as though one of them wins. They protect against different failures, and the reason the standard advice recommends both is not thoroughness for its own sake. It is that each one covers precisely the gap the other leaves.

Sorting out which failure you are actually worried about makes the decision straightforward.

The failures worth naming

Five things destroy data in ordinary life, and they are not equally likely.

  • Hardware failure. A drive stops. Common, and the reason most people buy a backup at all.
  • Deletion or overwriting. You delete the wrong folder, or save over a file, and notice three weeks later. Very common and often overlooked.
  • Theft or physical disaster. Burglary, fire, flood. Uncommon, but takes everything in the building at once, including the backup drive next to the computer.
  • Ransomware. Files are encrypted by malware, which deliberately looks for connected backups and encrypts those too.
  • Account loss. You lose access to the service holding the data, through a lockout, a billing lapse, or a provider closing.

Now put the two options against that list.

What a local backup does well

An external drive or a small network attached storage device on your own premises restores fast, which matters more than people expect. Recovering a full system from local storage is a matter of hours; the same recovery over a home internet connection can run for days.

It costs nothing after the hardware, has no ongoing subscription, and does not depend on a company continuing to exist or continuing to price its service the way it does today. For large volumes of data, photographs and video in particular, the cost difference over several years is substantial.

Where it fails: a drive sitting next to the computer is in the same building as the fire, the flood and the burglar. A drive left permanently connected is reachable by ransomware. And a single drive is itself a piece of hardware that will eventually fail, which is why a local backup that has never been tested is a guess rather than a backup.

What a cloud backup does well

Distance is the whole advantage. A copy in a data center hundreds of miles away survives everything that happens at your address, which covers the disaster category completely.

It also runs without being remembered, which is not a small thing. The most common failure of local backup is human: the drive was not plugged in for four months. A cloud service that syncs continuously does not depend on anyone doing anything.

Most services keep versions, which covers the deletion case. If you overwrite a file today and notice in three weeks, versioning is what gets the earlier copy back, and it is worth checking how far back a given service keeps versions before relying on it.

Where it fails: restore speed, ongoing cost, and dependence on an account. If the account holding the backup is the same account that gets compromised or locked, the backup goes with it, which is the argument for putting the backup somewhere with separate credentials and its own second factor.

Side by side

FailureLocal driveCloud service
Drive diesCoveredCovered
Accidental deletionOnly with versioningUsually covered
Fire, flood, theftNot coveredCovered
RansomwareCovered only if disconnectedCovered if versions are retained
Restore speedFastSlow for large volumes
Ongoing costNoneMonthly or annual

The combination that actually works

The old shorthand for this is three, two, one. Three copies of whatever you would hate to lose, held on at least two different kinds of media, with one of those copies somewhere other than the house. The rule predates cloud storage and translates cleanly: the working copy on your computer, a local drive for speed, and a cloud copy for distance.

Two refinements matter more than the rule itself. Disconnect the local drive between backups, because a drive that is always mounted is always reachable. And give the cloud backup its own credentials and its own second factor, separate from the email account you use for everything else.

For a household, the practical version is one external drive used weekly and kept in a drawer, plus one backup service running continuously. For a small business, add a copy kept somewhere physically different from the office, which for many people means a drive at home.

The step that makes any of it real

Restore something. Once a quarter, pick a real file from a month ago, restore it to a temporary location, and open it. This is the only evidence that any of the above works, and it is the step almost universally skipped.

Also check what is actually being captured. Backup software defaults commonly cover documents and pictures and skip application data, email stored locally, and anything on a second drive. The folder you assume is included is worth verifying once, and once is enough.

The Ready campaign run by the federal emergency management agency covers protecting documents and records as part of household disaster preparation, and its framing is the useful one: decide what you could not replace, then make sure a copy of it is somewhere the same event cannot reach.

About the author

Corinne writes for readers doing some of the work themselves.