That's why the redundancy design of disk arrays like IBM's XIV (and others) are so necessary with large SATA drives.
Rebuilding a single 8TB drive to another single 8TB drive in a raid array is dangerously slow (odds of a secondary failure are high). It will also have a long negative impact on array performance.
But if your rebuilding redundancy from a failed 8TB drive across a system containing say 72 of these drives with (at least) triple redundancy of any given extent, you'll be just fine and struggle to notice a performance hit.
Basically, large SATA drives don't belong in classic RAID designs. You want to massively distribute IO & redundancy to overcome the per disk bandwidth limitations.