I can not point you in the direction of a good strategy, but I can tell you that I have been on the receiving end of a bad one. By default B2D chunks your jobs into 1GB files, which can be changed to a significantly higher value. In this case, they were 1GB files.
The largest problems:
The array became very fragmented and could not be cleaned up unless all backup services were taken offline for an extended period of time (days).
The B2D files were backed up to tape, making the restore of purged or overwritten B2D files a two step processes (restore B2D file, then restore from B2D file).
Added confusion to log management.
Now I suspect the process could have been done better, but given the inherited frustration I trashed the B2D2Tape and am now using CPS, then backing up CPS job data to tape. System state is also pulled again when creating tapes (sent directly to tape).
Just one experience.