Home

Over the years several types of RAID have been developed. RAID stands for Redundant Array of Inexpensive Disks.  The basic reason is to allow for a large amount of storage on disk drives, but to be tolerant of disk failure without losing any data.  There is always a trade off between performance, cost and degree of possible failure.  There are several basic standards, but also many proprietary variations.  CnW Recovery concentrates on the most common standards with it's optional RAID recovery option.  Please contact CnW for details of purchase.  To evaluate the RAID option as part of the demo, enter the code 'RAID' in the registration code (rather than 'DEMO').

To use the RAID option there are two stages, the first is to configure the RAID, and the second then is to select the RAID and perform a recovery, just as with a standard drive.  ie a RAID will act logically in the same way as a single drive.

RAID recovery is often required when a RAID controller fails, or after an unsuccessful rebuild when a faulty drive is replaced.  The CnW Recovery tools will assist in all cases where data is still possible to be recovered.

The basic RAID standards that are in common use are as below

RAID 0

It is often argued that RAID 0 is not actually a RAID as it has no redundancy.  Data is striped between two or more disks.  It is method so that 2 500GB disks can be made to logically look like a 1TB disk.  With a suitable drive controller, both disks can be used at one time, and so can be faster than a single disk.  Failure of a single disk means that 50% of data is lost.  If a stripe is maybe 128KB long, then only some files, less than 128KB will be recoverable.  If there is a partial failure of one disk, then CnW Recovery software will produce good results.

RAID 1

RAID 1 is 100% redundancy.  The disk is imaged totally.  If one disk fails, the data is still on the other.  With a good hardware controller, performance will be the same as a normal disk.  With a software controller, speed will suffer.

RAID 4

RAID 4 is the same as RAID 5 below, except the parity stripe is always on a fixed drive, and not a moving location.  It is just as secure, but not always as fast.  For reading, the RAID 5 routines will work, but parity will be just one drive.

RAID 5

RAID 5 has at least 3 disks where one is a parity disk.  This will accept a single disk failure without losing any data.  Data is stored in stripes (maybe 128KB blocks at a time) and then the parity is calculated, and written on the final disk.  It is common for the parity to be written on different disks for different stripes and so for a 4 disk array, the data may be organised as below

       1   2   3   P

       4   5   P   6

       7   P   8   9

       P  10 11  12

For reading, performance is good, but when writing a single sector it is necessary to also read the other unchanged sectors on the same stripe, and then update the parity sector.  Thus what was a single write on a normal disk, becomes 2 reads and 2 writes.  This can be done in software with a device driver, but for high performance, a hardware controller is required.

RAID 6

RAID 6 is similar to RAID 5 except there are  two parity drives.  It means that recovery is possible even if two drives fail.  It can be used on 4 or more drives. The complex aspect is that the pattern used can be very varied.  For this reason on RAID-6 a CnW option of 'Variations' can be specified - this is when there is not single pattern but several.  (Only partially  implemented in CnW)

JBOD

Job Bunch of Disks.  This is the description given to multiple disks operating as a logical single drive.  This may or may not include redundancy.  One example is the HP media vault which has the string "Broadcom NAS Version 1.1 MBR Tag" as the start of disk 1.  This is normally a Reiser FS disk stored on several sections of multiple disks.  A specifc mode of JBOD has been added in CnW to handle these disks.

RAID Recovery

There are a few approaches to RAID recovery.  Depending on the application, and maybe forensic security, the disks may be read directly, or converted to a flat image to be processed in the same way as a single disk.  The software is very flexible on how disks may be read, either as a physical disk, or a an image file - even an image stored over a network.  A completely failed disk can be marked as missing

Variations

Variations is an option where the RAID configuration patter is longer than expected.  For instance, with a RAID-6, 4 drive one would expect 3 lines of configuration, such as

1  2  P  P

4  P  P  3

P  P  5  6

The possible problem with this configuration is that drive 1 and 4 could have twice the number of disk accesses than drives 2 and 3.  Thus the actual pattern found has been for 24 stripes, rather than 6 stripes.  To enter this with CnW it is necessary to set the variations value to 4, and fill in the data as follows (other RAIDS may have different configurations).

1  2  P  P

4  P  P  3

P  P  5  6

P  7  8  P

9  10 P  P

12 P  P  11

P  P  13 14

P  15 16 P

17 18 P  P

20 P  P  19

P  P  21 21

P  23 24 P

The variation value gives the number of variations - for most drives it will be 1, but the example above it was 4