Backup & Restore — Trivia & Interesting Facts¶
Surprising, historical, and little-known facts about backup and restore operations.
The "3-2-1 rule" was coined by a photographer, not a sysadmin¶
Peter Krogh, a digital photography expert, popularized the 3-2-1 backup rule (3 copies, 2 different media, 1 offsite) in his 2005 book "The DAM Book." The rule was so elegant that the entire IT industry adopted it wholesale, and it remains the bedrock of enterprise backup strategy two decades later.
Toy Story 2 was nearly lost to an accidental rm -rf¶
In 1998, someone at Pixar ran a variant of rm -rf * on the production filesystem for Toy Story 2, deleting 90% of the film. The backup system had been failing silently for weeks. The movie was saved only because Galyn Susman, the supervising technical director, had a copy on her home machine — she'd been working remotely while on maternity leave.
Magnetic tape is still the cheapest storage medium per terabyte¶
As of 2025, LTO-9 tape cartridges store 18 TB (45 TB compressed) for roughly $100, making tape about 5-10x cheaper per TB than the cheapest hard drives. IBM and Fujifilm demonstrated a tape prototype in 2020 storing 580 TB on a single cartridge. The tape industry quietly ships more total capacity each year than the HDD industry.
The world's first backup may have been punched cards from the 1890 census¶
Herman Hollerith's tabulating machines for the 1890 US Census used punched cards, and duplicate card decks were created as a rudimentary backup. The practice of duplicating card decks became standard — operators learned early that data loss was catastrophic when re-punching thousands of cards by hand.
Most companies discover their backups don't work only during a real disaster¶
A 2022 Veeam survey found that 58% of backup restores fail. The gap between "backup completed successfully" and "restore actually works" is so common it has a name in the industry: the "backup confidence gap." This is why restore testing — not backup testing — is the real metric.
The original rsync algorithm was a PhD thesis¶
Andrew Tridgell (also the creator of Samba) developed the rsync algorithm as part of his 1999 PhD thesis at the Australian National University. The rolling checksum technique he invented allows efficient delta transfers over slow links and remains essentially unchanged in modern rsync. His thesis is only 83 pages long.
GitLab accidentally deleted their production database in 2017¶
On January 31, 2017, a GitLab engineer ran rm -rf on the wrong database directory during a late-night maintenance window, deleting 300 GB of production data. Five different backup mechanisms had all failed or were misconfigured. They recovered most data from a 6-hour-old LVM snapshot that existed only by accident. GitLab live-streamed the recovery on YouTube.
Incremental forever backup was considered heretical when first proposed¶
Traditional backup schemes required periodic full backups (weekly full + daily incremental). When vendors first proposed "incremental forever" in the early 2000s — taking a single full backup and then only incrementals from that point on — backup administrators called it reckless. Today it's the default strategy for most modern backup products including Veeam, Commvault, and Cohesity.
The 1996 fire that created modern offsite backup awareness¶
On February 23, 1996, a fire destroyed the offices of the UK's Inland Revenue in London. Backup tapes were stored in the same building. The disaster became a landmark case study in business continuity planning and directly influenced the development of offsite and off-region backup regulations across the EU financial sector.
Amanda is the oldest open-source backup tool still actively maintained¶
Amanda (Advanced Maryland Automatic Network Disk Archiver) was created in 1991 at the University of Maryland by James da Silva. Over 30 years later, it's still in active development under Zmanda. It was designed to back up multiple machines to a single tape drive — a problem so fundamental that the tool outlived the technology it was built for.
Write-once-read-many (WORM) storage exists because of Arthur Andersen¶
After the Enron scandal in 2001 and the destruction of audit documents by Arthur Andersen, SEC Rule 17a-4 mandated that financial records be stored on non-rewritable, non-erasable media. This regulation single-handedly created the WORM storage market and later drove the development of S3 Object Lock and immutable backup features.
Deduplication ratios in backup can exceed 50:1¶
In environments with many similar VMs or containers, modern deduplication can achieve ratios of 50:1 or higher, meaning 50 TB of backup data compresses to 1 TB of actual storage. Data Domain (now Dell/EMC PowerProtect) popularized this with inline deduplication in 2004, and the technology was so effective that EMC acquired Data Domain for $2.4 billion in 2009.