Printable View
Hello,
I have read some docs explaining me each raid levels but I am still a bit confused between raid 1+o vs 0+1.I know oracle wants raid o+1 for systems/rdo/rbs but i couldnt figure the arguement between 1+0 vs 0+1. Can somebody in laymen's term explain me the difference??
thanks
Jigar
Hi Lakhani,
it is as simple as it.
RAID 0+1 : “Stripe First, Then Mirror What You Just Striped”.
RAID 1+0 : “Mirror First, Then Stripe Over What You Just Mirrored”.
Do you want some more details on it??
Hope it helps, All The Best
Sandy
RAID 1+0 is mirroring and then striping. RAID 0+1 is striping then mirroring.
For example, say you have four disks, sd0, sd1, sd2, sd3. RAID 1+0 would mirror disks sd0 to sd2 and sd1 to sd3. You would then create a logical volume striping the data across the two mirrored devices.
In RAID 0+1, you would create a stripe on sd0 and sd1, a stripe on sd2 & sd3, and create a logical mirror between the two striped volumes.
RAID 1+0 and 0+1 have similar performance. However, RAID 1+0 has more sustained redundancy. In the two examples above, lets assume sd1 goes bad. In the RAID 1+0, sd0 and sd2 are still mirrored. In the RAID 0+1, you are left with no redundancy.
Marist,
so if RAID 0+1 has no redundency then why does Oracle recommend it?
Whoa. RAID 0+1 does have redundancy. The redundancy is sacrificed when a disk goes bad which renders the stripe bad. Most arrays that have 0+1 capability have the capability to detect a bad drive, fail it, and substitute a hot spare. While the hot-spare is resyncing, you don't have any redundancy. Once the stripe is back to normal, you have your redundancy again. When a disk goes bad, you have mabye two hours where the stripe is vunerable.Quote:
Originally posted by jlakhani
Marist,
so if RAID 0+1 has no redundency then why does Oracle recommend it?
In a 1+0 system, up to 1/2 of your disks can go bad before you lose all your redundancy.