RAID with minimal (but really big) hard drives

Hi guys,

A thought occurred to me the other day, and I'm sure that it's 100% unoriginal and been discussed here a billion times, but I thought I'd post it anyway.

RAID X requires a certain number of hard drives (Y) where (in most cases) X = Y + 2 (e.g. RAID 5 = 3 HDDs + 2).

Theoretically, it be possible for someone to format a hard drive using a program like Partition Magic to create the number of HDDs needed (i.e. Y - 2). This would reduce the price of RAID considerably, seeing as a 1TB HDD is usually cheaper than 2 500GB HDDs. This is great for people who want 4 or more drives.

I'm wondering if this is actually possible, and has anyone done it?
4 answers Last reply
More about raid minimal hard drives
  1. It's possible and could be done using software RAID, but there would never be a reason to do it. All of the benefits of RAID go away, all of the problems with it come into play, and your performance would be abysmal.

    What you would be doing is taking what would be a single write to a single drive, dividing it into three or four, and spreading those writes out over as many different spots on the drive. i.e. you'd take your 70-to-100MB/s sequential read/write capacity, and you'd completely hose that in favor of your 5-to-10MB/s random read/write capacity.

    You'd also be getting rid of any redundancy benefits, because if that drive fails, all the partitions go with it. You never lose just one partition. You might get marginal protection against bad sectors, but at the expense of never being able to read or write faster than your worst-case scenario for those operations on that drive.

    Add to that the fact that you would have to do this in software RAID, so you would not be able to boot that drive (besides your boot time would be up by a factor of 5 or so), and you'd have the software layer riding on top taking a bunch of extra CPU time calculating the RAID striping.

    Now there IS a legitimate use of partitions in RAID, when you do use multiple disks. I recently had an array of 8 TB drives set up as both a Raid-0 and a Raid-10. I had about 15% partitioned at the beginning of each drive and the remaining 85% partitioned after that. The eight beginning partitions formed a 1.2TB raid-0 and the eight end partitions formed a 3.4TB raid-10.

    As it turned out, our disk usage was heavy on large reads, where Raid-0 and Raid-10 are pretty close to the same, and the whole thing was bottlenecked by the network, so the only thing I gained from the extra raid-0 was a bit more storage space. (4.6TB total, vs. the maximum 4TB if it was all in one.) We stored about 2.5 TB of data though so it didn't matter.
  2. Raid benefits come from reading/writing from multiple drives at the same time.

    Having extra partitions would not work, a hard drive doesnt have more then on head to read/write with so it could only handle one partition at a time.
  3. +1 what dashing guy said
  4. The benefit of a RAID array, and the principles on which it works, comes from the advantages you can gain by using several drives together, or more specifically, the advantage of using several read heads and several platters all at once.

    To vastly oversimplify, if 1 drive can transfer data at 100 m/ps, then 2 drives working together can transfer data at 200 m/ps. No matter how many partitions you put on a single drive, it cannot transfer data any faster, and access times actually go down. If they were done in a RAID type situation like you are talking about, access times would slow to a literally crawl because the single read head would have to jump from partition to partition to read a single file! That would be like using a hard drive that was somehow un-defragmented to the most extreme amount possible!

    Whoops, you guys are fast, or I am just too darn slow.
Ask a new question

Read More

NAS / RAID Hard Drives Storage