Sign in with
Sign up | Sign in
Your question

p5b:how many drives can die in raid 10 ? what about raid 01?

Last response: in Storage
Share
April 21, 2007 5:30:18 PM

well i have a p5b duluxe motherboard that has raid 10 ( it says it has that at least). I think the motherboard company made a mistake and placed raid 01 when it was supposed to be raid 10. aka the company lied.

my question: how many drives must fail in raid 10 to make the whole array fail?

how many drives must fail in raid 01 to make the whole array fail?
April 21, 2007 7:41:12 PM

The P5B Deluxe motherboard has 2 RAID controllers. One is the Intel ICH8 south bridge, part of the P965 chipset. It supports RAID 0, 1, 5, and 10. This is a true RAID 10, not a RAID 0+1. The other RAID controller is a 2-port JMicron controller, supporting RAID 0, 1, and JBOD.

In both RAID 10 and RAID 0+1, the number of drives that must fail to cause the entire array to be lost varies depending on which drives in the array fail. In all cases, a minimum of 2 failed drives can cause failure of the entire array. In the best failure case, up to 1/2 the drives in the array can fail and the array still be operational.

RAID 10 first pairs drives into RAID 1 groups, and then stripes across the RAID 1s. In RAID 10, you can theoretically lose 1/2 of the drives and not lose the array if all the failed drives are each one of the drives in their 2-drive RAID 1 group. If any entire RAID 1 group fails, the entire array is lost.

RAID 0+1 first stripes across 1/2 of the drives in the array to make a RAID 0. It then mirrors the entire RAID 0 to the other 1/2 of the drives in the array. You can theoretically lose 1/2 of the drives without losing the array if all the failed drives are confined to one RAID 0 group. If two drives fail where each is in a different RAID 0 group, the entire array is lost.

It's unlikely that you could experience two drive failures in RAID 0+1 that isolate both failures to the same RAID 0 group. The reason for this is that when the first drive goes down, the entire RAID 0 group it was in is lost. Thus, the other drives in that RAID 0 group won't even be operational. The only operating drives still working in the array are in the other RAID 0 group, making the 2nd failure more likely to occur there, which will then lose the entire array.

This is not the case for RAID 10. A single drive failure makes that particular RAID 1 pair non-redundant, but only the remaining operating drive in that RAID 1 pair is the single point of failure. Any other operating drive in other RAID 1 groups can fail and the array remains intact.

RAID 10 also outperforms RAID 0+1 during the rebuild operation following a drive failure and replacement. In RAID 0+1, all drives in the entire array have to participate in the rebuild operation (since the rebuild re-mirrors one RAID 0 to the other). By contrast, in RAID 10, only the other drive in the RAID 1 pair participates in the rebuild; other drives in the array have nothing to do with it.
April 21, 2007 9:25:22 PM

Quote:
The P5B Deluxe motherboard has 2 RAID controllers. One is the Intel ICH8 south bridge, part of the P965 chipset. It supports RAID 0, 1, 5, and 10. This is a true RAID 10, not a RAID 0+1. The other RAID controller is a 2-port JMicron controller, supporting RAID 0, 1, and JBOD.

In both RAID 10 and RAID 0+1, the number of drives that must fail to cause the entire array to be lost varies depending on which drives in the array fail. In all cases, a minimum of 2 failed drives can cause failure of the entire array. In the best failure case, up to 1/2 the drives in the array can fail and the array still be operational.

RAID 10 first pairs drives into RAID 1 groups, and then stripes across the RAID 1s. In RAID 10, you can theoretically lose 1/2 of the drives and not lose the array if all the failed drives are each one of the drives in their 2-drive RAID 1 group. If any entire RAID 1 group fails, the entire array is lost.

RAID 0+1 first stripes across 1/2 of the drives in the array to make a RAID 0. It then mirrors the entire RAID 0 to the other 1/2 of the drives in the array. You can theoretically lose 1/2 of the drives without losing the array if all the failed drives are confined to one RAID 0 group. If two drives fail where each is in a different RAID 0 group, the entire array is lost.

It's unlikely that you could experience two drive failures in RAID 0+1 that isolate both failures to the same RAID 0 group. The reason for this is that when the first drive goes down, the entire RAID 0 group it was in is lost. Thus, the other drives in that RAID 0 group won't even be operational. The only operating drives still working in the array are in the other RAID 0 group, making the 2nd failure more likely to occur there, which will then lose the entire array.

This is not the case for RAID 10. A single drive failure makes that particular RAID 1 pair non-redundant, but only the remaining operating drive in that RAID 1 pair is the single point of failure. Any other operating drive in other RAID 1 groups can fail and the array remains intact.

RAID 10 also outperforms RAID 0+1 during the rebuild operation following a drive failure and replacement. In RAID 0+1, all drives in the entire array have to participate in the rebuild operation (since the rebuild re-mirrors one RAID 0 to the other). By contrast, in RAID 10, only the other drive in the RAID 1 pair participates in the rebuild; other drives in the array have nothing to do with it.


how do i actually know if its raid 10 ( true raid 10) ? when i boot up it says

raid 10 ( 0+1 )

which confuses me...

i will take a picture of it soon.
April 21, 2007 10:05:41 PM

you can loose up to 2 drives in either a 10 or 0+1, in a raid the there are 2 mirrors that are stripped AA and BB you can loose one drive from the A mirror and one drive from the B mirror. In a raid 0+1 there are two stripes that are mirrored AB and AB and you can loose either set of AB.
!