RAID 0 with 3 different size drives except only 2 theoretical drives

bigmacsa

Prominent
Dec 5, 2017
10
0
510
I searched google and all I could find was the answers about raid 0 being limited to the smallest size so 100GB, 250GB, 500GB, and 1,000GB in raid 0 array being limited to 400GB because of the smallest drive being 100GB and it cant fit any more of the 1/4 striped data onto it.

But my question is about using dual-striped data across 3 drives, 2 of which are just "stacked"

250+250=500
+
500
=
1,000

here is a visual i created
https://imgur.com/a/hl7KYI2

It would basically be the same exact thing as a normal raid 0 with "2 drives" except one of the "drives" is actually a larger partition made of 2 smaller drives that are not in raid 0.

and since the 2 smaller drives wouldn't be in raid they technically don't even have to be the same size. so another example could be

100+400=500
+
500
=
1,000

I can't think of why it wouldn't be possible. If it's not can you provide an answer as to why?
 
RAID 0 wants to write data across all volumes at once. That is why the minimum size is required. If you want just a concatenated JBOD, you could do what you want. The first disk would be written, then when the first disk fills, then next one, and so on. You won't get RAID 0 performance. You will get single disk performance with a volume size like you describe.
 
A RAID 0 depends on the size of the smallest drive in the array. That's the way it is, no way around that.
Whatever combination you can devise...the size of the smallest drive dictates the eventual size of the whole array.

Could you, in theory, make some weird RAID 0 combination of ((250_250)+500) = 1TB?
Yes, probably. And massively increase the complexity and fail potential, for a performance benefit that is easily beaten by any standard SSD.

Any other combination of those drives is just a JBOD, resulting in normal HDD performance, and the only thing you sort of gain is having all those in a single volume and drive letter.

Have fun with a backup routine for that conglomeration.

Windows and applications have gotten really good with managing individual drives and multiple drive letters. ADding in the probable age of those 100/250GB drives, and the ever reducing price of drive space, makes it even less of a good idea.
 
thank you for the responses but my brain still isn't satisfied. i'm not sure if it's because i understand too much or too little. the answer "raid just wants to write to all drives at the same time and there's no way around it" just doesn't sound right or maybe isn't elaborate enough. if raid data is striped across 2 disks, whats to stop it from viewing 2x250GB disks as 1x500GB disk and write half of the striped data to it as a virtual 500GB+500GB raid 0 array?
In my mind the benefits/limitations would be basically twice the speed of the slowest drive since it would literally be striping to 2 disks at any given time and not all 3. it will only start writing to the second of the 2 smaller drives when the striped data fills up the first smaller drive.

here is a visual i created
https://imgur.com/a/hl7KYI2

i understand it would introduce another failure point but it should theoretically be less of a failure chance than buying 2 more 250GBs to put in 4x raid 0 correct? it could give upgrade options to people like me who want to increase their storage without doubling failure chance on top of the already doubled failure chance.. I have exactly this situation. I have 2x250GB SSD's in raid 0 but I could take them out of raid and stack them as a single volume, then buy a new 500GB SSD and increase the storage to 1TB while keeping 2x raid 0 performance and not having to deal with separate disks and manage where things are stored if they were JBOD.

hopefully someone can give me either a more detailed explanation as to why this is impossible or maybe a solution.

thanks again for your answers
 
All of it ends up in one RAID 0.

250+250+RAID 0 = 500GB array
500GB array + 500GB HDD + RAID 0= 1TB array

Failure of any drive, or burp of the RAID controller = all data is lost across the whole system.
RAID 0 writes the stripe across all drives at the same time.

For performance that is completely crushed by a $130GB SSD.


For your current 2x250GB SSD in RAID 0?
http://www.tomshardware.com/reviews/ssd-raid-benchmark,3485.html
Zero user facing performance benefit.