Software
Controlled RAID Solutions
Software
controlled RAID is simply a core component of an operating system,
or an application that handles construction and administration of
a RAID array attached to the system by conventional means such as
multiple hard drives connected to a SCSI card. This type of RAID
system provides the lowest form of RAID control and does not usually
provide any additional EDAP features.
Many network operating systems such as NT and Linux for example,
provide a limited software RAID ability within the OS itself to
build server-dependent arrays.
The RAID software component built into NT Server has the ability
to create either a Level 0 stripe set without
parity, a Level 1 mirrored pair, or a Level
5 stripe set with parity. The data is striped across the disks
in 64K blocks much the same as the default setting of most controllers.
As with dedicated controllers and RAID implementations, the minimum
requirement is two physical hard drives - you cannot use two partitions
as all sections of the stripe set must be accessed simultaneously.
These arrays obviously rely completely on the stability of the server
they are attached to and in no manner offer any redundant mechanisms
to cover server failure. Since NT Server implements volume and stripe
sets in software, NT must be running in order to recognise and access
them.
Advantages:
Software based RAID is the most inexpensive way of entering the
RAID market. RAID software simulates a dedicated RAID controller
by running on top of the operating system and allowing RAID arrays
to be created and assigned to the system without the requirement
for any other hardware.
Disadvantages:
This type of solution however carries slight penalties in performance:
a) striping drives on a system-based SCSI chain will dramatically
increase the I/O performance when compared to stand alone drive
subsystem, b) the CPU of the machine is used by the software to
actually drive the RAID set-up, reducing system performance with
I/O, parity, and resource overhead.
Most
software based solutions utilise the speed of a SCSI bus and drives
to achieve the optimum performance. A typical solution would consist
of the software itself, a SCSI host controller, and SCSI drives.
The RAID subsystem is also at the mercy of the operating system
and machine it is attached to. If the host fails then the RAID subsystem
fails. Although the data residing on the subsystem will be intact
and will no longer be available until either the original host is
reconfigured and put back on-line, or an alternative host is configured
to take it's place.
Related
topics:
|