Outside of the SQL Server world, deduplication of data in backups is big business. It is the way that large organisations are making $M savings on storage costs.

But this is hard to do with SQL Server backups, unless the backup is done by a backup agent. But most SQL Server folk do not like backup agents because historically they are unreliable when it comes round to doing a recovery.

The lack of ability to deduplicate the data in the SQL backups is maybe a major reason behind the proposal to just do a quarterly backup. In theory you could do daily differential backups to a quarterly full backup, but the management of this amount of backup to me looks horrendous.

We need Microsoft and other backup vendors to come up with a solution that allows deduplication of SQL backup data, otherwise the storage costs for unduplicated SQL Backups will start to become a reason for avoiding SQL Server.

An idea that is used by another DBMS is called 'offline backup merge'. This takes a full backup and a differential backup (or a chain of log backups) as input, and produces a new full backup showing the state as at the end of the merged backup. This can be done on any server, with the merge process connecting to the original DB instance at the end of the work to register the new backup. This process is not the full data deduplication that is possible with block-level backups, but it can help simplify database recovery for slowly changing data.