Our schema runs a full back up every 14 days, and then goes back to incemental... each user is limited to 3gb of disk space, so the over head is not too bad.... we purge the old back ups that are 45 days old, this limits the server side overhead to to about 20gb per user with archived back ups, and since the users are all remote users on dsl & cable we have throttled the full backup to 125/Ksec, and by staggering the date of full back up for the users the total load to the network is at a minimum.
We are in the process of moving to the DLO agent for all of our remote users, this should lessen the overhead even more!!