Never Restored a Backup? The Test That Turns a Claim Into a Fact

A backup you have never restored from is a claim about the future, not a record of anything. Most people setting one up for the first time buy storage, watch a progress bar finish, and treat the green checkmark as the end of the job. The checkmark reports that files were copied. It does not report that they can be read back, in a usable order, onto a machine you do not yet own, on a day when you are already having a bad one. The distance between those two statements is the whole subject, and it is measurable.

The three copies came from people who had lost two

Rotation schemes predate personal computers. Mainframe operators ran daily, weekly, and monthly tape sets, a scheme usually called grandfather-father-son, because a single tape reused every night gives you exactly one bad moment before the only good copy is overwritten by a corrupted one. The familiar modern shorthand, three copies of the data on two kinds of media with one copy off site, was popularized by the photographer Peter Krogh in his 2009 book on digital asset management, written for people whose livelihood sat on external drives. Each element of the rule answers a failure someone actually had: drive death, media format failure, and the house fire or theft that takes both drives sitting on the same desk.

RAID and sync solved different problems, and got mistaken for this one

The 1987 Berkeley paper that gave us the term RAID was about keeping a server available through a disk failure, not about keeping your history. A mirrored pair writes your mistake to both disks at once, faithfully and instantly. Consumer file sync arrived roughly two decades later and inherited the same confusion in a friendlier package: a folder that appears on every device, a deletion that appears on every device just as fast, and a ransomware encryption that appears on every device faster still. Sync is a distribution system. It becomes a recovery system only where the provider keeps versions, and only as far back as the retention window you are paying for, which is a number worth reading before you rely on it.

Ransomware turned the annual restore drill into a monthly one

The crypto-extortion wave that began in earnest around 2013 changed what a backup has to survive. Before it, the threat model was accident: a failed drive, a deleted folder, a flood. After it, the threat model included an attacker who looks for your backup target first, because a reachable backup is the thing standing between them and payment. The industry response was immutability, meaning copies that cannot be altered or deleted for a fixed period even by an account with full credentials, alongside the older habit of keeping one copy fully offline. The National Institute of Standards and Technology is responsible for the cybersecurity framework that most American organizations now map their recovery practices against, and recovery there is a named function, not an afterthought bolted to storage.

What a first restore test actually measures

Run three tests, in increasing size, and time each one with something other than your impression. Restore a single file from thirty days ago and confirm it opens in the application that made it. Restore a folder of several gigabytes and check the file count and the modification dates, not just the folder size. Then restore a full system image to spare hardware or a virtual machine, which is the only test that reveals missing drivers, an encryption key you never wrote down, or a boot partition your tool quietly skipped. The number you want out of this is hours, and for most first-timers it is larger than the guess by a factor that surprises them.

Which dimensions set the price

Storage capacity is the figure quoted and rarely the figure that decides the bill. Retention length is the first real driver: keeping ninety days of daily versions of a working set is a different product from keeping one current mirror. Restore behavior is the second, since some providers charge for the data you pull back out, or ship a physical drive for a fee, or throttle a large recovery in a way that turns two hours into two days. The third is machine count, because per-device pricing punishes households with five laptops and rewards a single server. Measure your working set in gigabytes, your acceptable data loss in hours, and your acceptable downtime in hours, and the right tier usually declares itself.

Write the three numbers down where the backup software cannot see them, on paper or in another account, along with the encryption key and the account recovery email. The first restore you perform under pressure should be your second or third restore overall.