Andrew, thanks for the insight. I know the differential is all changed data since last full backup. I just put this schedule in place and will monitor the size of the diff backups, if they become too big, I will change the full backup from weekly to daily.

The production DB is currently about 220GB, and the current schedule has taken full backups each day, with four hour incrementals. The incrementals go from 0.5GB (12AM), 1.5GB (4AM), 2.5GB (8AM), 4.5GB (12PM) to 5.2GB (4PM) in size (yesterday, 8PM was full backup time, so no differential scheduled then). It looks like a 1 GB increment (roughly) every four hours (6 GB/day) - maybe a little less, since some of the recent data will be modified several times over the first few days of it's life - production, QA testing, costing, shipping.

Compunding the differentials (42 in a week) will yield about 900 GB worth of backup data plus 220 GB for the full backup. The previous schedule would carry four full sets of backup of 220 GB for the DB and 15 GB for the differentials which is about the same storage, but would take less backup time overall, or so I am predicting. I might also be able to change the differential window to six hours, roughly cutting the storage space of differential logs to two thirds (from 42 compounded to 28 in a week). But this will depend on restore time.

Of course, I may be totally off on any of these counts and will be adjusting the schedule within a week of setting it up 🙂

Edit: These are uncompressed figures, but the logic should be unaffected.