Sequential Data Striping - Hitachi Universal Storage Platform V Reference Manual

Track 0
to
Track 7
Track 32
to
Track 39
next
8
tracks
Parity
Tracks
Figure 3-2
RAID-6. A RAID-6 array group consists of eight (6D+2P) data drives. The
data is written across the eight drives in a stripe that has six data chunks and
two parity chunks. Each chunk contains either eight logical tracks (mainframe)
or 768 logical blocks (open).
In the case of RAID-6, data can be assured when up to two drives in an array
group fail. Therefore, RAID-6 is the most reliable of the RAID levels.

Sequential Data Striping

The Universal Storage Platform V/VM's enhanced RAID-5+ implementation
attempts to keep write data in cache until parity can be generated without
referencing old parity or data. This capability to write entire data stripes, which
is usually achieved only in sequential processing environments, minimizes the
write penalty incurred by standard RAID-5 implementations. The device data
and parity tracks are mapped to specific physical drive locations within each
array group. Therefore, each track of an LDEV occupies the same relative
physical location within each array group in the storage system.
In a RAID-6 (dual parity) configuration, two parity drives are used to prevent
loss of data in the unlikely event of a second failure during a rebuild of a
previous failure.
3-4
Hitachi Universal Storage Platform V/VM User and Reference Guide
RAID-5 using 3D + 1P and 3390-x LDEVs
Track 8
to
Track 15
Track 40
to
Track 47
Parity
Tracks
next
8
tracks
Sample RAID-5 3D + 1P Layout (Data Plus Parity Stripe)
Functional and Operational Characteristics
Track 16
Parity
to
Tracks
Track 23
Track 24
Parity
to
Tracks
Track 31
next next
8 8
tracks tracks
next next
8 8
tracks tracks