Overland Storage SnapSAN S1000 User Manual page 125

SnapSAN S1000 User Guide
• RAID 5 - highly available storage
RAID 5 is similar to RAID 0 in that data is striped across multiple disks. However, one
disks worth of space is reserved to store parity data, which can be used to reconstruct
the pool in the event of one of its disks failing. With RAID 5, the parity data is
distributed across all the disks in the pool. If a single disk fails, each block of data
stored on that disk can be reconstructed using the corresponding data block from all
the other disks along with the parity block. This means that if a single disk fails, data
can still be read, albeit at a rather slower rate (because it needs to be reconstructed,
rather than read directly). For this reason, a RAID 5 pool with a disk failure is referred
to as a degraded pool.
With a RAID 5 pool, because data is read from many disks in parallel, as for RAID 0,
read performance is good. Write performance is slightly lower because, in addition to
writing the data, parity data has to be calculated and written. If a hardware RAID
controller is used, this will be done using dedicated hardware; if software RAID is
used, the work will be done on the main processor of the storage controller.
The capacity of a RAID 5 pool is reduced by exactly one disks worth of capacity, which
is required to store the parity data. For example, a RAID 5 pool made up of 4 x 100 GB
disks will have a capacity of 300 GB.
In principle, a RAID 5 pool could have a very large number of disks. However, the more
disks there are, the greater the chance of a double disk failure. If a single disk fails, the
data is no longer protected until the disk has been replaced and the pool has been
rebuilt by reconstructing all the data from the failed disk and writing it to the new
disk. If the disk capacities are very large, it may take many hours to rebuild the pool.
If a second disk fails before the rebuild has completed, all the data in the pool will be
lost. That is to say, large capacity disks increase the time taken to rebuild the pool,
during which time the pool is vulnerable to a second disk failure. Moreover, the chance
of a second disk failure increases as the number of disks in the pool increases.
• RAID 6 - very highly available storage
RAID 6 is similar to RAID 5 but instead of storing a single disk's worth of parity data,
two disk's worth are stored, making the pool capable of withstanding the failure of two
disks. However, there is an additional write overhead involved in calculating the
double parity data. Since RAID 6 works best with dedicated hardware, RAID 6 is only
offered on systems with a hardware RAID controller. Read performance is similar to
that of RAID 0 or 5. Since two disks are used for storing parity data, the capacity of a
RAID 6 pool made up of 8 x 100 GB disks will be 600 GB.
• RAID Level 10
RAID 10 is defined as mirrored stripe sets or also known as RAID 0+1. You can build
RAID 10 either directly through the RAID controller (depending on the controller) or
by combining software mirroring and controller striping, or vice versa (called RAID
01).
• RAID Level 50
A RAID 50 combines the straight block-level striping of RAID 0 with the distributed
single parity of RAID 5. That is, a RAID 0 array striped across RAID 5 elements. It
requires at least 6 disks. This can increase the performance by allowing the controller
to more efficiently cluster commands together. Fault tolerance is also increased, as one
drive can fail in each individual array.
• RAID Level 60
A RAID 60 combines the straight block-level striping of RAID 0 with the distributed
double parity of RAID 6. That is, a RAID 0 array striped across RAID 6 elements. It
requires at least 8 disks. This can increase the performance by allowing the controller
to more efficiently cluster commands together. Fault tolerance is also increased, as two
drives can fail in each individual array.
10400310-003 01/2013
©2010-13 Overland Storage, Inc.
Glossary
GL-10
