Hi Srinivas,
>>>>> "Srinivas" == srinivasa rao <srinivas> writes:
Srinivas> Hi all I am having a little bit of confussing in
Srinivas> raid(Redundant array of Inexpensive disks). The basic
Srinivas> idea of RAID was to combine multiple small, inexpensive
Srinivas> disk drives into an array of disk drives which yields
Srinivas> performance exceeding that of a Single Large Expensive
Srinivas> Drive.In Linux how internally it will do?. Can I call
Srinivas> this as hardware cluster.Can any one plz help me
Srinivas> regarding this.
The four popular levels of RAID are:
* RAID-0, also known as Striping. This is not strictly RAID, since
all it does is permit you to create a logical device that spans
multiple partitions, possibly on multiple spindles. Results in the
ability to create filesystems larger than physical disk size, and
possibly in faster data throughput on concurrent access. Does not
survive disk crashes.
* RAID-1, aka Mirroring. Requires two identical partitions, usually
on separate disks, which are combined into one logical device. Data
written to the RAID device is written to both of the partitions. If
one disk goes bad a copy of the data is still available on the other
one. Survives single-disk crashes, requires 100% overhead (since you
require 1MB+1MB on each disk to write 1MB of data).
Performance is slightly slower than single disk for writing, may be
faster on a properly-designed RAID-1 system.
* RAID-3, aka Parity Disk. Requires n+1 disks for n disks worth of
data. All n+1 disks are identical. Data written to the RAID device
is split into n chunks and each chunk is written to a data disk. A
CRC (error-correcting checksum) of the data is written onto the n+1'th
disk, also called the parity disk.
If any of the first n disks goes bad the data in the remaining disk
plus the CRC in the parity disk can be used to reconstruct the data.
If the parity disk goes bad the CRC can be reconstructed from the data
in the first n disks. Survives single-disk crashes, requires
100*n/(n+1) percent overhead (e.g. if you use 4 data disks, you need
one parity disk, i.e. 25% overhead).
Slow for writing since the parity disk becomes a bottleneck; fast for
reading.
RAID-5, aka Striped Parity. Like RAID-3, only there is no special
parity disk. The CRC is striped over the data disks along with the
data. Requires n+1 disks for n disks worth of data, so overheads are
identical to RAID-3. Reasonably fast for both reading and writing.
Hope that helps,
-- Raju
--
Raj Mathur [EMAIL PROTECTED] http://kandalaya.org/
It is the mind that moves
================================================
To unsubscribe, send email to [EMAIL PROTECTED] with unsubscribe in subject
header. Check archives at http://www.mail-archive.com/ilugd%40wpaa.org
