HP Surestore NAS - Manual (Page 77)

77
Also, if a third disk in the LUN fails while in a degraded state, parity can no longer be used
and all data in the LUN becomes inaccessible.
Figure 3 is an example showing the distribution of user data and parity data in a RAID 5DP
configuration. The example shows one LUN with five stripes, each stripe with five segments:
three data segments and two parity segments (P n and Q n ). The segments are striped across
the disks in a rotating fashion. The following data protection scheme is demonstrated for RAID
5DP:
o Any one disk can fail without losing data redundancy. If one disk fails, the data
on that disk can be recalculated using either the P or Q parity segment and the data
stored on the remaining good segments in the stripe.
o Any two disks can fail without losing data. If two disks fail simultaneously (B and
C, for example), or if a second disk fails before the first disk is completely rebuilt, the
data from both failed disks can be rebuilt. For example, both segments of missing
data in stripe 1 on disk B and disk C can be recalculated using the P 1 and Q 1 parity
data and the remaining data segment on disk A. Alternatively, if the segment on a
lost disk is either P or Q parity data, then the parity data can be recalculated from the
remaining user data.
Note: Disks A and E are considered as adjacent disks. The segment size is some number (x)
times 520-byte blocks.
Figure 3: RAID 5DP
RAID 1+0 and RAID 5DP Combined (AutoRAID)
Your Virtual Array uses a combination of RAID 1+0 and RAID 5DP (or AutoRAID) to store data.
This means that the same disk within a redundancy group can have one portion of its data
capacity stored in RAID 1+0, while the other portion is stored in RAID 5DP. If one disk fails, the
array first rebuilds the data in RAID 1+0 (the RAID level most exposed to data loss), then rebuilds
the data in RAID 5DP. On the other hand, if two disks fail simultaneously, the array first rebuilds
the data in RAID 5DP, then rebuilds the data in RAID 1+0. Thus, your array optimizes the data
rebuild process, greatly reducing your exposure to data loss.
Your virtual array will survive two simultaneous disk failures within a redundancy group, without
losing data, as long as none of the data segments in the disks are mirrors of each other.