Backup Size Calculator

Backup sizing is a capacity problem driven by retained copies rather than by the source data itself. The full-backup term is the protected volume reduced by the compression ratio, while the incremental term is the protected volume multiplied by the daily change rate and then reduced by the same compression factor. The count of retained full images equals the retention window divided by the full-backup interval, rounded up, which is why a 30-day retention with weekly fulls holds roughly five complete images instead of one. Daily change rate, often called churn, is the dominant unknown: general file servers typically run 1 to 5 percent per day, transactional databases 2 to 10 percent, and continuously recording video surveillance approaches 100 percent. Compression behaves just as unevenly, with text, logs and database dumps giving 60 to 80 percent reduction while JPEG, MP4 and already-zipped content give almost none. The model assumes a constant churn rate and a uniform compression factor across the whole data set, so treat its output as a planning envelope rather than a measurement. Real products change the arithmetic in ways this formula does not capture: block-level change tracking makes incrementals smaller than file-level tracking, synthetic full backups avoid re-reading the source entirely, and global deduplication across copies can cut retained storage by another factor of 10 to 50 in environments full of near-identical virtual machines. Typical results land between 3x and 6x the source volume. Retention is often set by rule rather than by cost, with HIPAA requiring 6 years and SOX 7 years for the relevant record classes.

Systematic backup practice grew out of magnetic tape operations in the 1950s, beginning with the UNISERVO drive on the UNIVAC I in 1951 and the IBM 726 in 1952. Because tape was expensive and slow, operators developed rotation schemes rather than simply writing everything every night. The Grandfather-Father-Son scheme, which keeps daily, weekly and monthly generations at different retention lengths, and the Towers of Hanoi rotation, which spaces reuse of each tape geometrically, both date from this mainframe era and still describe how most retention policies are shaped today. The explicit full-versus-incremental distinction was formalized in Unix by the dump utility in the 1970s, which introduced numbered dump levels where level 0 is a full image and each higher level captures only what changed since the most recent lower-level dump. Commercial products generalized the idea through the 1990s: IBM ADSM, released in 1993 and later renamed Tivoli Storage Manager, popularized progressive incremental or incremental forever backup, in which a single full is followed indefinitely by incrementals that the server reassembles on demand. Deduplicating disk targets arrived in the 2000s with Data Domain, founded in 2001, shifting the economics away from tape. The widely quoted 3-2-1 rule of three copies, two media types and one offsite copy was popularized by photographer Peter Krogh in The DAM Book in 2005 and has since become standard guidance well beyond photography.