raid 3 drives?

With a True Hardware RAID Controller ...
It will be 2 active drives with the 3rd drive in hot standby, should one of the active drives fail the RAID set will reconfigure the RAID set to the hot standy drive and the remaining active drive. All the drive should be allmost the same size as they all will be formatted to the same same by the controller.

Eh? I can only assume you're talking RAID 1 here ... the rest of us are talking RAID 0. If you lose a drive in RAID 0 your data is gone. You cannot configure a hot-spare for a RAID 0 array.
 
masticore, you are wrong. I have a raid 3 system. 2 striped 1 parity. it is done through an intel matrix raid controller.

see http://www.prepressure.com/techno/raid.htm

quote:
"On RAID 3 systems, datablocks are subdivided (striped) and written in parallel on two or more drives. An additional drive stores parity information. You need at least 3 disks for a RAID 3 array".
 
masticore, you are wrong. I have a raid 3 system. 2 striped 1 parity. it is done through an intel matrix raid controller.
So you don't have a RAID 0 system ... what's your point? I never said you can't use 3 drives in a RAID 3 array (you need at least 3 drives for RAID 3). I said "You cannot configure a hot-spare for a RAID 0 array" ... where's the reference to RAID 3 ????

see http://www.prepressure.com/techno/raid.htm

quote:
"On RAID 3 systems, datablocks are subdivided (striped) and written in parallel on two or more drives. An additional drive stores parity information. You need at least 3 disks for a RAID 3 array".
yes, thanks. I know what RAID 3 is. Do you know that different levels of RAID exist and ... GOSH! omg ... they have different requirements and capabilities and advantages and disadvantages? Amazing stuff hey .....
 
Sorry!

I misunderstood you.

but you can have performance and reliablility in a RAID0 by addeding another drive to be a hot backup.

then of course its no longer a raid0 but a raid3.

however most onboard types like the silicon image and ones round in nforce3/4 only let you do a 0 or 1. mirror or stiping. so if you have one of these boards then its either performance or reliability. raid 3 or raid 10 would be the compromise of the two systems.

sorry boet!
 
I misunderstood you.
I assumed as much ;)

but you can have performance and reliablility in a RAID0 by addeding another drive to be a hot backup.

then of course its no longer a raid0 but a raid3.
Not really. You cannot add a hot-spare to a RAID 0 array. What you're talking about is adding a dedicated drive for parity - not the same as a hot-spare. In any event, RAID 3 and 4 are mostly dead nowadays. The de facto standard if you're using 3 drives is to go with RAID 5; the pros outweigh the cons.

however most onboard types like the silicon image and ones round in nforce3/4 only let you do a 0 or 1. mirror or stiping. so if you have one of these boards then its either performance or reliability. raid 3 or raid 10 would be the compromise of the two systems.
Some new, advanced boards are providing RAID 5. As I said, 3 and 4 are dead. But ja, RAID 10 is still an option and it's better than 0+1.

sorry boet!
no problem. Re-reading my response I think I was a bit too blunt. Apologies :D
 
my dfi nforce 4sli expert has RAID 5.........is raid 5 basically 4 hard drives, basically 2xRAID 0's ; with the second pair copying everuthing from the first as backup?
 
my dfi nforce 4sli expert has RAID 5.........is raid 5 basically 4 hard drives, basically 2xRAID 0's ; with the second pair copying everuthing from the first as backup?

no. What you've described is RAID 0+1. RAID 5 expands on RAID 3 and 4.

RAID 3 and 4 (and subsequently RAID 5 as well) require at least 3 drives to operate. In RAID 3 and 4, one of the drives is allocated as a dedicated parity drive. Data is written, striped, across all except the parity drive. A "checksum" is then calculated for the written stripe which is then written to the parity drive. A technically inaccurate (but sufficient to get the concept across) example would be this-

if the data "ABCDEF" was to be written to the array, and the array was made from 4 drives, then data would be written as such:

drive 1: A
drive 2: B
drive 3: C
drive 4 (parity drive): checksum of A, B and C
drive 1: D
drive 2: E
drive 3: F
drive 4 (parity drive): checksum of D, E and F

The purpose of the parity drive is to permit the array to continue running should one of the drives fail. So if drive 2 in the above example failed, the actual data that was written to drive 2 could be calculated from the available data (first stripe: A, C and the parity; second stripe D, F and the parity.) The RAID controller then presents the "real" data to the OS.

RAID 5 took 3 and 4 further by removing the dedicated parity drive. In RAID 5, parity is distributed evenly across all the drives. So something like this happens (using the same example data and number of drives):

drive 1: A
drive 2: checksum of A, B and C
drive 3: B
drive 4: C
drive 1: D
drive 2: E
drive 3: checksum of D, E and F
drive 4: F

In all of RAID 3, 4 and 5 you lose the capacity of 1 drive due to parity being stored. So if you had 4 x 100GB drives in a RAID 3, 4 or 5 setup, your available space would only be 300GB (100GB being "lost" to store the parity info.)

Also, most (decent) RAID controllers don't require that the drives are the same sizes, BUT the controller will only work across "partitions" (not the same as an operating system partition, but the correct term eludes me right now) of equivalent sizes. So if you have 3 x 100GB drives and 1 x 80GB drive, you'll only be able to get the storage space of 3 x 80GB drives (only 80GB of each of the 100GB drives will be used).
 
Top
Sign up to the MyBroadband newsletter
X