Disk Enclosures
1753265 Members
5386 Online
108792 Solutions
New Discussion юеВ

step by step instructions for 0+1

 
Nathan Stipe
Occasional Advisor

step by step instructions for 0+1

I have no problem understanding the theory to this raid configuration. I am about to purchase a raid controller and would just like to be more clear about how to set this up when the time comes.

I am looking for step by step instructions on how to do this within the ACU. I know I make two seperate raid 0 arrays. What I would like to know is how I tell them to mirror within ACU?

More important is this question. I need to know if this is possible. If I want to add capacity to this setup (in equal amounts to each raid 0 of course), will the precedures described in the 532 documentation for volume expansion with dynamic disks work? Has anyone out there successfully added capacity to an 0+1 under win 2k with dynamic disks, or with basic disks as well?

Thanks
2 REPLIES 2
Chris Vail
Honored Contributor

Re: step by step instructions for 0+1

I can't help you with the Win2k question(s), but the theory is the same regardless of the operating system.
RAID0 is mirroring. This means that you need 2 drives of the same size. This gives you the equivalent of ONE disk. So (2) 40GB disks yields only a TOTAL of 40GB. This one is easy: the RAID controller merely sends all write information to both disks. Reading usually comes from a primary, and if that fails, the secondary.
RAID 1 is striping. This one is a little harder to understand. In striping, sequential data is written to sequential drives, in turn. For example: your OS sends 30 "Blocks" (or 512 byte segments) of data to your striped volume set. You've set up your RAID 1 with 5 disk drives, without parity. This means that blocks 1,6,11,16,21,and 26 are written to the first disk. Blocks 2,7,12,17,22 and 27 are written to the second disk. Blocks 3,8,13,18,23, and 28 are written to the 3rd disk. Blocks 4,9,14,19,24,29 are written to the 4th disk. And finally blocks 5,10,15,20,25 and 30 are written to the 5th disk. Consquently, when this data is read back in, a single read FROM ALL 5 DISKS SIMOULTANEOUSLY, you get blocks 1-5 at the same moment, blocks 6-10 in the next, blocks 11-15 next, then 16-20, 21-25, and 26-30. Instead of making 30 read operations, the computer has made 6. This quintuples the apparent speed of the disk drive system.
As a practical matter, RAID1 striping doesn't buy much until you get 3 or more disks in a stripe set. Our systems here have 10 disks per stripe set, yielding ~10 times the speed of a single disk drive (its not exactly linear, but close).
In order to have RAID 0+1, you have to mirror your stripe sets. This means, as a minimum, you have 6 drives to play with, which wiil yield a disk size three times as fast, and three times as large as a single disk. But it will be mirrored for reliability.
RAID 5 (for your education) is advanced striping. With this, you need an additional disk drive to your RAID set. As a rule, RAID 5's are only practical with 5 or more disks. The 5th disk contains redundant, or parity information. You probably know that normally, there are 8 bits in a byte. Parity adds a 9th bit, so that if any of the other 8 bits are lost, the data can be reconstructed from the 9th bit. RAID 5 takes that alorhythm a bit furter, with the parity block on the extra disk. Some controllers break up the parity data amongst all disks in a stripe set. This means that if any single disk is lost, not all parity data is lost along with it. In any case, RAID 5 is superior (in some ways) to RAID 0 in that you get more efficient use of your disk drive resources. With 10 drives in a RAID 5, you would get the use of 9 of those disks. With the same 10 disks in a RAID 0 configuration, only 5 will be useful. Less efficient of all would be a RAID 5+0, which is mirroring, striping and parity all at the same time. Out of the 10 drives that you start with, only 4 will be useful, but then you'd have parity AND mirroring to protect your data.
I hope this helps

Chris
Nathan Stipe
Occasional Advisor

Re: step by step instructions for 0+1

I had said in my post that I understand all the theory. What I am wanting to know is how 0+1 is accomplished specifically within the "ARRAY CONFIGURATION UTILITY".

I know I create two striped arrays and then mirror them against each other. Is this the terminology used in the ACU? Is it possible in ACU to tell two arrays to mirror each other? If so, this would be stupid though, it seems it would be possible to create two raid 5 arrays and mirror them against each other?