Hey Aiten,
I suspect our collective silence is down to not many of us trying this!
We looked into this a little bit a while back, and the only option that didnt come with IO penalties that would have been too steep (LUKS encypted filesystem, IIRC) was to use encypted-at-rest drives (this was HP at the time), but after investigation it was decided that the additonal cost wasn't worth the additonal protection; as we mostly store genomic data on RAID array filesystems, so you would need multiple drives from the RAID to reconstruct a portion of the file which was itself compressed (CRAM for e.g.) and so the possibility of recovering usefull data from a recyled drive wasn't considered high enough to warranty the additional expense (and vendor lock in!).
I'm pretty sure the community would be interested in any experimentation you might have done though?
John