I noticed something very similar on regular firewalls before as well, specially comparing size of scheduled backups after upgrade. Oddly enough, I noticed same difference in backup size in cluster config, though that could be expected, as only master would process traffic, so thats probably normal.
I really cannot say why upgrade would cause that issue, as I had never seen it back in R77 and before that days, but in R80, maybe since database is way different, not really sure.
Might be worth asking TAC about it...would be nice to get official answer.