The Antares 5070 is a high performance, versatile, yet relatively inexpensive
host based RAID controller. Its embedded operating system (K9 kernel) is modelled
on the Plan 9 operating system whose design is discussed in several papers from
AT&T (see the "Further Reading" section). K9 is a kernel
targeted at embedded controllers of small to medium complexity (e.g. ISDN-ethernet
bridges, RAID controllers, etc). It supports multiple lightweight processes
(i.e. without memory management) on a single CPU with a non-preemptive scheduler.
Device driver architecture is based on Plan 9 (and Unix SVR4) streams. Concurrency
control mechanisms include semaphores and signals.
The 5070 has three single ended ultra 1 SCSI channels and two onboard serial
interfaces one of which provides command line access via a connected serial
terminal or modem. The other is used to upgrade the firmware. The command line
is robust, implementing many of the essential Unix commands (e.g. dd, ls, cat,
etc.) and a scaled down Bourne shell for scripting. The Unix command set is
augmented with RAID specific configuration commands and scripts. In addition
to the command line interface an ASCII text based GUI is provided to permit
easy configuration of level 0, 1, 3, 4, and 5 RAIDs.
Text based GUI for easy configuration for all supported RAID levels.
A Multidisk RAID volume appears as an individual SCSI drive to the operating
system and can be managed with the standard utilities (fdisk, mkfs, fsck,etc.).
RAID Volumes may be assigned to different SCSI IDs or the same SCSI IDs but
different LUNs.
No special RAID drivers required for the host operating system.
Multiple RAID volumes of different levels can be mixed among the drives forming
the physical plant. For example in a hypothetical drive plant consisting of
9 drives:
2 drives form a level 3 RAID assigned to SCSI ID 5, LUN 0
2 drives form a level 0 RAID assigned to SCSI ID 5, LUN 1
5 drives form a level 5 RAID assigned to SCSI ID 6, LUN 0
Three single ended SCSI channels which can accommodate 6 drives each (18 drives
total).
Two serial interfaces. The first permits configuration/control/monitoring of
the RAID from a local serial terminal. The second serial port is used to upload
new programming into the 5070 (using PPP and TFTP).
Robust Unix-like command line and NVRAM based file system.
Configurable ASCII SCSI communication channel for passing commands to the 5070's
command line interpreter. Allows programming running on host OS to directly
configure/control/monitor all parameters of the 5070.
Much of the information/knowledge pertaining to RAID levels in this section
is adapted from the software-raid-HOWTO by Linas Vepstas . See the acknowledgements
section for the URL where the full document may be obtained.
RAID is an acronym for "Redundant Array of Inexpensive Disks"
and is used to create large, reliable disk storage systems out of individual
hard disk drives. There are two basic ways of implementing a RAID, software
or hardware. The main advantage of a software RAID is low cost. However, since
the OS of the host system must manage the RAID directly there is a substantial
penalty in performance. Furthermore if the RAID is also the boot device, a drive
failure could prove disastrous since the operating system and utility software
needed to perform the recovery is located on the RAID. The primary advantages
of hardware RAID is performance and improved reliability. Since all RAID operations
are handled by a dedicated CPU on the controller, the host system's CPU is never
bothered with RAID related tasks. In fact the host OS is completely oblivious
to the fact that its SCSI drives are really virtual RAID drives. When a drive
fails on the 5070 it can be replaced on-the-fly with a drive from the spares
pool and its data reconstructed without the host's OS ever knowing anything
has happened.
The different RAID levels have different performance, redundancy, storage capacity,
reliability and cost characteristics. Most, but not all levels of RAID offer
redundancy against drive failure. There are many different levels of RAID which
have been defined by various vendors and researchers. The following describes
the first 7 RAID levels in the context of the Antares 5070 hardware RAID implementation.
RAID-linear is a simple concatenation of drives to create a larger virtual drive.
It is handy if you have a number small drives, and wish to create a single,
large drive. This concatenation offers no redundancy, and in fact decreases
the overall reliability: if any one drive fails, the combined drive will fail.
SUMMARY
Enables construction of a large virtual drive from a number of smaller drives
No protection, less reliable than a single drive
RAID 0 is a better choice due to better I/O performance
Also referred to as "mirroring". Two (or more) drives, all
of the same size, each store an exact copy of all data, disk-block by disk-block.
Mirroring gives strong protection against drive failure: if one drive fails,
there is another with the an exact copy of the same data. Mirroring can also
help improve performance in I/O-laden systems, as read requests can be divided
up between several drives. Unfortunately, mirroring is also one of the least
efficient in terms of storage: two mirrored drives can store no more data than
a single drive.
SUMMARY
Good read/write performance
Inefficient use of storage space (half the total space available for data)
RAID 6 may be a better choice due to better I/O performance.
Striping is the underlying concept behind all of the other RAID levels. A stripe
is a contiguous sequence of disk blocks. A stripe may be as short as a single
disk block, or may consist of thousands. The RAID drivers split up their component
drives into stripes; the different RAID levels differ in how they organize the
stripes, and what data they put in them. The interplay between the size of the
stripes, the typical size of files in the file system, and their location on
the drive is what determines the overall performance of the RAID subsystem.
Similar to RAID-linear, except that the component drives are divided into stripes
and then interleaved. Like RAID-linear, the result is a single larger virtual
drive. Also like RAID-linear, it offers no redundancy, and therefore decreases
overall reliability: a single drive failure will knock out the whole thing.
However, the 5070 hardware RAID 0 is the fastest of any of the schemes listed
here.
SUMMARY:
Use RAID 0 to combine smaller drives into one large virtual drive.
Best Read/Write performance of all the schemes listed here.
No protection from drive failure.
ADVICE: Buy very reliable hard disk drives if you plan to use this scheme.
RAID-2 is seldom used anymore, and to some degree has been made obsolete by
modern hard disk technology. RAID-2 is similar to RAID-4, but stores ECC information
instead of parity. Since all modern disk drives incorporate ECC under the covers,
this offers little additional protection. RAID-2 can offer greater data consistency
if power is lost during a write; however, battery backup and a clean shutdown
can offer the same benefits. RAID-3 is similar to RAID-4, except that it uses
the smallest possible stripe size.
SUMMARY
RAID 2 is largely obsolete
Use RAID 3 to combine separate drives together into one large virtual drive.
RAID-4 interleaves stripes like RAID-0, but it requires an additional drive
to store parity information. The parity is used to offer redundancy: if any
one of the drives fail, the data on the remaining drives can be used to reconstruct
the data that was on the failed drive. Given N data disks, and one parity disk,
the parity stripe is computed by taking one stripe from each of the data disks,
and XOR'ing them together. Thus, the storage capacity of a an (N+1)-disk RAID-4
array is N, which is a lot better than mirroring (N+1) drives, and is almost
as good as a RAID-0 setup for large N. Note that for N=1, where there is one
data disk, and one parity disk, RAID-4 is a lot like mirroring, in that each
of the two disks is a copy of each other. However, RAID-4 does NOT offer the
read-performance of mirroring, and offers considerably degraded write performance.
In brief, this is because updating the parity requires a read of the old parity,
before the new parity can be calculated and written out. In an environment with
lots of writes, the parity disk can become a bottleneck, as each write must
access the parity disk.
SUMMARY
Similar to RAID 0
Protection against single drive failure.
Poorer I/O performance than RAID 3
Less of the combined storage space is available for data [than RAID 3] since
an additional drive is needed for parity information.
RAID-5 avoids the write-bottleneck of RAID-4 by alternately storing the parity
stripe on each of the drives. However, write performance is still not as good
as for mirroring, as the parity stripe must still be read and XOR'ed before
it is written. Read performance is also not as good as it is for mirroring,
as, after all, there is only one copy of the data, not two or more. RAID-5's
principle advantage over mirroring is that it offers redundancy and protection
against single-drive failure, while offering far more storage capacity when
used with three or more drives.
SUMMARY
Use RAID 5 if you need to make the best use of your available storage space
while gaining protection against single drive failure.
