Configuring the RAID Array (Redundant Array of Independent Drives) means taking more
than three hard drives and making one large partition. Sounds simple doesn't it?
As with most server services and hardware it sounds simple but really isn't
until you understand the process.
This is a hardware solution! Not to be confused
with Windows Disk Management!
There are three main types of RAID configurations: RAID 0, RAID 1, and RAID
5. We will not consider the other types in between. (This
applies to the newer SATA and IDE controllers that are add on cards or built in
to the motherboard).
RAID 0 is when you have a RAID controller in a system and don't use the RAID
facility of the controller.
RAID 1 is one of the most often used to backup the OS drive.
RAID 5 is the other most often used configuration for enlarging and making a
redundant partition.
RAID 1 also known as 'Mirroring' is when you have your OS on one drive and
make an exact copy of another drive. When the mirror process is complete any
thing that is written to the original drive will also be written to the mirror
drive. This is a excellent way to backup the OS drive because you can write all
security and configuration to the mirror drive with out any extra software or
security considerations.
RAID 5 configuration allows for you to replace one failed hard drive with out
losing any data. This is very important when dealing with very large volumes of
data. Be aware that RAID 5 DOES NOT take the
place of scheduled backups! The way it works is the controller will write data
across all but one of the drives, then on the last drive it will write a parity
bit with the location of the written data, you can select the size of each part
of data that is written, 64 bytes is the normal size. A visualization: you have
a RAID controller in your server, other than the OS drive you have four more
drives, say 36 gig each. When you setup the RAID 5 the controller will combine
three of the drives into one partition of 108 gig (approximately) and the fourth
drive will become the parity drive. So you have lost 36 gig of drive space! Yes,
but if you have one of the drives fail all you have to do is replace the drive
and your data will be safe!
Ok, now you know why you should use a RAID controller and configure it in
your server.
But Wait! There's More!
What more can there be? I mean you need a RAID controller, you need to
'Mirror' the OS drive, and use RAID 5 on the data drives. Well there are a
couple of things you should know, some pit falls of the different RAID
configurations. Do's and Don'ts of using the RAID controller.
Considerations - SCSI or SATA?
SCSI (Small Computer Serial Interface) is the most common drive and
controller interface at this time. First introduced in the mid 1980's it has
been the main stay for servers because of the data transfer speed and
reliability. The controllers have built in memory and cache, if power fails the
data will be retained by the back up battery for the cache on the controller
card. Although the controllers and drives are moderate to expensive (depending
on the size of the drive of course).
SATA (Serial Advanced Technology Attachment) is the 'new kid on the block'
and is coming into it's own. Originally only for workstations and only one or
two drives per system now there are RAID controllers that can accommodate up to
16 drives and soon more. A SATA controller and drives would be comparatively
cheaper than the traditional SCSI configuration. (Check this page if
installing SATA or IDE drives).
Tid bits that you may want to know -
When you get your RAID controller you will receive a cd or floppy with the
current drivers on it, safe guard this data! Make yourself a bootable device
(floppy, usb, or cd) with the drivers on it incase of catastrophic failure.
When you have a hard drive failure replace it right away, if you have two
failures the data on the RAID will be lost. The
RAID controller can only recover from ONE drive failure
not multiple drive failures.
As the server ages you should also consider powering it down when there are
long periods of inactivity. A server normally runs 24/7 never being shut down
for any longer than it takes to replace a failed part. Because the materials of
the computer expand with heat they do not get a long cool down period. This
continuous expansion will over time take its toll on the soldier joints and pcb
boards of the components. To over come this if the computer is powered off once
a week for a period long enough to dissipate the heat build up and return to
their pre heat expansion state you will have less failures especially hard
drives, next to the processor a hard drive will create the most heat inside the
computer.
The last tid bit for the RAID configuration you should be aware of is this:
When you make a RAID 5 with SCSI controller and drives you should use all the
same size drives. What the controller sees is the smallest drive you want to
create the array with and resize all the other drives to that size. The reason
for this is the parity, the parity bit is dependent on the overall size of the
drive, if you have four drives, three 36 gig and one 16 gig then the array would
be 48 gig with one parity drive, you will lose 48 gig!
In other words use the same size and type of drive when you use
RAID 5 solution.
After making your array with the hardware controller you need to make the volume
with Windows Disk Management, which has it's own idiosyncrasies.
This site contains a lot of information. As
with any publication not all information is available due to space, time, or
subject constraints.
If you have a question that you did not find the answer
on this web site you a can
ask your question here and we will endeavor to get you the most up to date
answer possible!
process.
most heat inside the
computer.
