andrew gothard (8/17/2016)

---

Very briefly:

Full -

allows you to restore to a point in time - say 4 seconds past 5 this afternoon when someone accidentally truncated the X table in live instead of Dev.

if you're taking your t-log backups remotely (remotely can mean to dedicated backup storage or a remote datacentre) you can define a Recovery Point Objective. Ie, this is the Max amount of minutes of data I can lose if I lose a box / database. If you take them every 5 minutes, in principle you're talking about a max loss of 5 minutes worth of data

This, you'd generally want for OLTP

Simple

no point in time restore - only from the last full or differential backup. All changes since then are lost.

Common use-case would be reporting databases after your nightly ETL process