For me, it depends on the data loss policies of the company, and the ability to recover data.
For example, I used to manage a database that was just a tool for managing and processing mailing lists. Since the lists themselves existed outside the database, and could be re-imported and re-processed pretty easily, and since the database itself was huge (terrabytes), I didn't bother with anything beyond a weekly full backup and daily diff backups.
For another example, I used to manage a database with business-critical financial data, where the loss of any data at all could have catastrophic consequences to the business. For that, I kept the most recent two full backups, and the log backups for those, on disk, and kept a full history of full backups off-site in a vault (that last was partially due to legal requirements). With that, and the backups being tested right after they were taken (both checksum and actual restore), I could be pretty certain I could recover to at least the most recent log backup. And since backups, logs, and data files were on separate SANS, I could be pretty certain I could deal even with a full disk cratering on one of the SANS without data loss.
Most of what I've worked with has been in between those.
So, it depends on the business need.
- Gus "GSquared", RSVP, OODA, MAP, NMVP, FAQ, SAT, SQL, DNA, RNA, UOI, IOU, AM, PM, AD, BC, BCE, USA, UN, CF, ROFL, LOL, ETC
Property of The Thread
"Nobody knows the age of the human race, but everyone agrees it's old enough to know better." - Anon