How to use this database backup policy guide
Enter the full backup cycle, the intermediate interval and the generations you keep, and the page reports restore point counts, the worst-case data loss window and the steady-state storage total. It proposes no cycle or retention period of its own.
How it is calculated
Full backups inside the daily window are ceil(retention days / full cycle), and intermediate backups per cycle are ceil(full cycle x 24 / interval) minus one. One incremental is the daily change volume times interval over 24, while a differential accumulates after the full, so the jth one is j times that size. Compression multiplies by the saving you enter. As of October 2026 no recognised recommended cycle or retention period exists, so every figure stays an input.
Limits and cautions
Real storage shifts with deduplication, the share of changed blocks, transaction logs and the snapshot technology in use. These are steady-state figures that assume a constant daily change volume, so growth rate and disk exhaustion are not reported. Designing backwards from recovery objectives belongs to a different tool.
Frequently asked questions
There is no recognised recommended cycle. Tolerable data loss, storage cost and restore time all differ by service. This page proposes no schedule; it computes the loss window and storage a schedule you enter would produce so you can compare options.
An incremental backup holds only what changed since the previous backup, while a differential holds everything changed since the last full backup. Differentials grow over the cycle but restore from one full plus one differential, whereas incrementals stay small and need the full plus every incremental in between.
A failure right after a backup loses almost nothing, but a failure just before the next one loses the whole interval. Capacity planning has to assume the worst case, so the interval is used directly as the loss window.