Originally Posted by HybridCore;13690661
...snip...What do you mean?
In RAID 0+1 or RAID 1+0 (also referred to as RAID 10), the two types or RAID work together to bring you better performance and fault tolerance (they both have different results. Check out Manyak's guide
to RAID). However, Intel Rapid Storage Technology doesn't make the two work together. It splits the drive(s) into two parts, one for RAID x and one for RAID y. It allows you to store important data on the RAID array that has higher fault tolerance and unimportant data on the RAID array that has better performance.
I think the process should go something like this...
...Where's the RAID 0?
You might be wondering where the RAID 0 is in my example of what I think it should be like. The RAID 0 occurs between the 2 discs. Since both drives hold all of the data, the fault tolerance is 1. No data is lost if one drive is damaged. As most of us know, RAID 0 works by using 2 or more separate discs to retrieve parts of the data. It takes twice as much (or more, depending on how many drives you use) time to get the complete file from one drive than two drives getting 2 different parts that make up the complete file. What happens here is that since both drives have part a and b. The first drive retrieves part and the second drive retrieves part b. That's where the RAID 0 comes in.
I understand the first bits of what you're saying, and I know how Matrix works. But it seems you aren't clear about what RAID10 (or RAID0+1) are, how they work and what the requirements are. And your listed 'cons' for Matrix aren't really cons of Matrix itself, but are actually down to physical limits on the technology (it's not physically possible to do a 'RAID0' with mechanical drives as you seem to indicate in your post, and have both performance and redundancy together).
Now I may be getting the wrong impression of what you're trying to say though, which is why I asked for extra info...
Yep, just seen your new diagram in the OP. What you have shown is RAID1, and this can't be used as RAID0 and be expected to give the same speed boost. It just isn't physically possible with mechanical drives, which rely on the rotation of the platters in the drive and are limited by how long this takes.
With SSDs however you can get much closer to what you describe - and guess what... Matrix does it already, and does it pretty well. And it's likely that AMD, NVida, LSI, Areca, Adaptec, etc do so too. They are all just using RAID1, but will split the reads across different disks. With mechanical drives doing sequential transfers it doesn't actually help thoough...Edited by the_beast - 5/31/11 at 4:32am