As pointed out in Section I briefly, this even distribution of
parity chunks in RAID-5 could cause the correlated multiple
failure problem because it leads that all SSDs are worn out at
similar times. To avoid this, Diff-RAID allocates more parity
chunks to an older SSD to differentiate the aging rate of
SSDs. Figure 1(b) shows how Diff-RAID allocate parity chunk
among SSDs. For example, when three different pages, A0, B1,
and C2, are randomly written to the disks, Disk 0 gets three
writes while the others gets single write.