As far as I know, migration is processed as follows:

ADSM migrates by nodes. At first, the node with the largest amount of
storage in the disk pool, then second, etc etc. Each node one process.

If this is true, it sounds like you have lots of nodes with only few
files backed up each. In this case you could think about move data of
your disk pool (as a circumvention)?

Regards

Roger