There has always been issues and debates about full and incremental in the IT world since the inception of "backups".
First, let me explain the problem.
As you know, a full backup will take everything and put it somewhere.
An incremental will only take the changes that it knows about from the last time it did any kind of backup, whether it was a full or an incremental.
Now, here's where the real problem begins, as if the above sentance wasn't enough of one already.
If you have a corrupted backup of any of the incrementals between the full and the time of recovery, you're screwed. You can't recover from an incremental that was partial; therefor the data is now inconsistant from last known good state.
If you're doing CRC and other checksum checks on the data WHILE it's being recorded.. and you store the media in a safe, protected place, AND you know where it is to find it.. well.. incrementals MIGHT work. But my experience is that no matter how secure you think you are in making sure the backups were successful.. media can change from the time the job is done, until the time the tape/media is moved from the device to the storage area.
Do a FULL backup of production every day.
Don't use the same media more than 3 times.
I use RAID 1+0 on all production devices.
I do FULL Backups every day, DISK to DISK (MUCH FASTER) from production servers to SAN storage arrays, and then I backup the SAN storage devices to not interfere with performance from production.
Once a month do a restore from your backup situation to a fake prod server and test functionality.
Hope this helps..