The different raid levels available today
Raid 0 - Stripping data across the disks.
This stripes the data across all the disks present in the
array. This improves the read and write performance. Eg. Reading a large file takes a
long time in comparison to reading the same file from a Raid 0 system.They is no data
redundancy in this case.
Raid 1 - Mirroring.
In case of Raid 0 it was observed that there was no redundancy,i.e if one
disk fails then the data is lost. Raid 1 overcomes that problem by mirroring the data. So
if one disk fails the data is still accessible through the other disk.
Raid 2
RAID level that does not use one or more of the "standard" techniques of mirroring,
striping and/or parity. It is implemented by splitting data at bit level and spreading it
across the data disks and redundant disk. It uses a special algorithm called as ECC
(error correction code) which is accompanied across each data block. These are tallied
when the data is read from the disk to maintain data integrity.
Raid 3 - data is striped across multiple disks at a byte level.
The data is stripped with parity and
the parity is maintained in a separate disk. So if that disk goes off , it results in a data
loss.
Raid 4
Similar to Raid 3 the only difference is that the data is striped across multiple disks at
block level.
Raid 5
Block-level striping with distributed parity. The data and parity is stripped across all
disks thus increasing the data redundancy. Minimum three disks are required and if
any one disk goes off the data is still secure.
Raid 6
Block-level striping with dual distributed parity. Its stripes blocks of data and parity
across all disks in the Raid except that it maintains two sets of parity information for
each parcel of data thus increasing the data redundancy. So if two disk go off the data
is still intact.
Raid 7
Asynchronous, cached striping with dedicated parity. This level is not a open industry
standard. It is based on the concepts of Raid 3 and 4 and a great deal of cache is
included across multiple levels. Also there is a specialised real time processor to
manage the array asynchronously.
Raid 0 - Stripping data across the disks.
This stripes the data across all the disks present in the
array. This improves the read and write performance. Eg. Reading a large file takes a
long time in comparison to reading the same file from a Raid 0 system.They is no data
redundancy in this case.
Raid 1 - Mirroring.
In case of Raid 0 it was observed that there was no redundancy,i.e if one
disk fails then the data is lost. Raid 1 overcomes that problem by mirroring the data. So
if one disk fails the data is still accessible through the other disk.
Raid 2
RAID level that does not use one or more of the "standard" techniques of mirroring,
striping and/or parity. It is implemented by splitting data at bit level and spreading it
across the data disks and redundant disk. It uses a special algorithm called as ECC
(error correction code) which is accompanied across each data block. These are tallied
when the data is read from the disk to maintain data integrity.
Raid 3 - data is striped across multiple disks at a byte level.
The data is stripped with parity and
the parity is maintained in a separate disk. So if that disk goes off , it results in a data
loss.
Raid 4
Similar to Raid 3 the only difference is that the data is striped across multiple disks at
block level.
Raid 5
Block-level striping with distributed parity. The data and parity is stripped across all
disks thus increasing the data redundancy. Minimum three disks are required and if
any one disk goes off the data is still secure.
Raid 6
Block-level striping with dual distributed parity. Its stripes blocks of data and parity
across all disks in the Raid except that it maintains two sets of parity information for
each parcel of data thus increasing the data redundancy. So if two disk go off the data
is still intact.
Raid 7
Asynchronous, cached striping with dedicated parity. This level is not a open industry
standard. It is based on the concepts of Raid 3 and 4 and a great deal of cache is
included across multiple levels. Also there is a specialised real time processor to
manage the array asynchronously.
