I FLAC them, and store them on my ZFS-based
OpenIndiana (a Solaris derivative) file server. This is currently set up in a RAID-Z2 configuration (roughly equivalent to RAID-6, with two redundant drives). In turn, the storage pool gets copied to a separate set of disks periodically, also ZFS. ZFS is extra-paranoid, going so far as checksumming every block written, and never overwriting live data. I've configured the system to perform a "scrub" (checking for bad blocks) every week. Since the motherboard is a recent AMD-based ASUS, it supports (and I use) ECC RAM for extra reliability.
The main storage pool is currently 3 TB, with 8x500 GB Western Digital Blue drives, hanging off an Intel SASUC8I controller. The motherboard has six SATA ports, and there's also a two-port SATA card as well, in addition to the SASUC8I. One solid-state drive (now a Samsung, after the previous OCZ failed) provides caching. Losing the cache drive doesn't cause data loss.
I'm considering the idea of switching the main storage to two mirrored pairs of 1.5T drives, and leaving the remaining four slots open for further expansion. All drives except the SSD are in hot-swap cages; if necessary I could offline a failed HD, yank it out, and slam in a new one without shutting down.