On 23 Feb 2001 09:00:10 +0100, Danilo Godec wrote:
> Is there a point in having a RAID-5 array made of 3 (or more) RAID-1
> arrays?
>
> md0 = sda1+sdb1
> md1 = sdc1+sdd1
> md2 = sde1+sdf1
>
> And then, finaly, a RAID-5 of those:
>
> md3 = md0+md1+md2
Sure, there's a point to this. If you lose three disks all at once,
then you're still up and running. If you lose a fourth disk, you may or
may not be up. 5 disks out and you're screwed. This is going to give
you very high reliability, with smaller disk space loss than from other
methods. This method is still expensive, as you cut down useable disk
space by 1/2, then cut it down by 1/3 of that. What sort of scenarios
are you looking at this for?
> Or a RAID-1 array made of two RAID-5 arrays?
Nope, the other arrangement makes more sense when you think about
failure cases.
Which kernel version do you want to use. It's my understanding that
this is possible using the RAID code in 2.4, but as of yet I've not
tried it. (I'm waiting for RH to ship 2.4 first). Actually, you should
be able to layer any raid layers that you want to with 2.4, from what
I've been told. :-)
Greg
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to [EMAIL PROTECTED]
