HP StorageWorks 1500 Modular Smart Array installation guide (355901 - 005, June 2007)

Installation procedures—All deployments26
• Vertical striping—Offers ultimate fault tolerance and performance, but at the price of storage
efficiency.
• Horizontal striping—Allows for the creation of large arrays and efficient use of storage capacity,
but at the price of I/O performance and less fault tolerance.
In a vertical configuration, an array uses hard drives from separate storage enclosures and different
SCSI buses. In a horizontal configuration, the array uses multiple drives contained within the same
storage enclosure.
RAID levels
In conjunction with the striping method, the assigned RAID level also determines the fault-tolerance,
I/O performance, and storage efficiency of the LUN.
See Table 4 below for a comparison list of the different RAID levels.
When selecting a RAID level for a LUN, consider the type of data that will be stored on the array.
For example:
• For transitional data: you may want to use RAID 0, which provides no fault tolerance, but
provides rapid storage and access of large amounts of data.
• For critical data: use a fault-tolerant RAID level such as RAID 1, RAID 1+0, RAID 5, or RAID 6.
Choose the RAID level that offers the desired combination of fault-tolerance, I/O performance,
and storage efficiency.
Depending on the assigned RAID level, one or more drives within a LUN can fail without bringing
the drive sub-system down.
Table 4 RAID level comparison
RAID level Alternative name I/O
performance
Fault tolerance Storage
efficiency
RAID 0 Data striping Highest None Highest
RAID 1
RAID 1+0
Drive mirroring
Data striping plus drive mirroring
High * Highest * Low
RAID 5 Data striping, with one set of
distributed parity data
Medium Medium High
RAID 6 Data striping, with two sets of
distributed parity data
Low High Medium
* Drives in the array need to be striped across separate enclosures on different SCSI buses.When drives are
selected for an array, the controller always mirrors the first half of the drives selected to the second half of the
drives selected. Therefore, if the drives selected are on a different SCSI bus or in a different enclosure, the
maximum fault tolerance levels will be realized.