One Formatted Drive, Decades of Backup Lessons

Aleksandar Filipovski tells a simple story about a formatted drive and turns it into a masterclass in backup planning. The piece, titled “Backups Aren’t Simple,” has been making the rounds on Hacker News, and it deserves a second look beyond the laugh it earns.

Filipovski’s dad plugged an external hard drive into a TV set-top box and was prompted to format it. The photos stored there were lost, the index of files gone. The drive looked empty, but the data was dead. The lesson was cheap: never keep everything in one place.

The One Drive Problem

The story opens with a quote Filipovski heard years ago. “There are two types of people: those who have suffered a catastrophic loss of data, and those who will.” He confirms he belongs to the first group. His family had moved all their photos onto one external drive to free up laptop space. When the drive was formatted, the files were gone.

It would be easy to blame Filipovski’s father. But the piece argues that the mistake was systemic. They hadn’t backed up the drive. Consumer software usually warns against formatting a disk, but the set-top box did not. Why should a non-technical person have to know that?

The photos were restored eventually. But the point stuck: drives die, they’re stolen, and even cold storage can rot data. Hard drives shift their magnetic particles. SSDs leak electricity and corrupt files over time. The risk is real, and it isn’t rare.

The Backup Rules

The core rule is simple. Have a copy of your files somewhere else. That’s the first principle.

But a plugged-in drive can still fail. Ransomware could encrypt it. A script could overwrite everything with zeroes. A mirror backup — something like RAID — isn’t enough. You need snapshots, not a mirror.

That brings in the Recovery Point Objective, or RPO. Critical financial institutions might demand a recovery point measured in seconds. Small businesses might tolerate a day or more. Filipovski suggests a weekly backup for photos, meaning six days and 23 hours of data could be lost.

Rotating Snapshots

Snapshots pile up fast. A daily backup with a one-day RPO creates 365 snapshots a year. That’s a lot of storage.

Filipovski proposes a rotation scheme. Keep recent snapshots frequent — daily backups rotated every 14 days, weekly backups rotated every 7 weeks, monthly backups rotated every year. The granularity changes over time. What happened on day 2 of the year matters less when it’s day 364.

This is called a GFS-rotated, snapshot-based backup. The list of adjectives grows as the plan gets more complex.

Deduplication and rsnapshot

Then comes the compression insight. Video frames in movies are mostly the same, with only a few pixels changing. Files follow a similar pattern. Most files don’t change at all. A tiny minority change constantly.

That led to deduplication. Instead of storing identical copies, you store one file and reference it from each snapshot. This survives backup rotation because you delete directory entries, not files themselves. rsnapshot uses this approach, storing only the changes between two adjacent snapshots — an incremental backup.

The savings are real.

What This Means for You

The piece is a reminder that backup is a discipline, not a purchase. It requires planning, rotation, and attention. It also requires accepting that some data loss is inevitable, and that the goal is to minimize it, not eliminate it entirely.

Filipovski’s journey from formatted drive to RPO is a useful one. It moves from personal tragedy to technical practice without losing the human scale. The story is funny, but the advice is sound.

Here’s the plan broken down:

• Take daily snapshots and rotate them every 14 days
• Take weekly snapshots and rotate them every 7 weeks
• Take monthly snapshots and rotate them every year

The takeaway is straightforward: copy your files somewhere else, take snapshots, and rotate them. The details — RPO, GFS-rotated, rsnapshot — are for the curious. The core problem remains the same as it was in that living room decades ago: one drive, one failure, and no plan.

It’s a cheap lesson in handling data. You never keep important things in one place only.

backup · data loss · deduplication · rsnapshot · storage