Stripe RAID using X25-M gen 1 + X25-M gen 2?

[aibo]

Hi guys,
Would there be any performance penalty for using these two different drives in a RAID 0? I wouldn't expect the stripe to go as fast as using two gen 2 X25-M's, of course. But would it at least be as fast as if I were using two gen 1 X25-M's?

I wanna do a stripe RAID and it doesn't really make sense to buy 1 more old X25-M model to match.

 

[nanofrog]

...But would it at least be as fast as if I were using two gen 1 X25-M's?

It should , but I've not tested SSD's in RAID (any level).

You might want to consider getting another Gen 1 though, as the prices will hopefully fall below the Gen 2's.

 

[Tesselator]

Theoretically it would be a combination of the speeds - so faster than two gen-1 units but not as fast as two gen-2 units. Theoretically. But I can't think of anything that would negate the hypothetical.




.

 

[nanofrog]

Theoretically it would be a combination of the speeds - so faster than two gen-1 units but not as fast as two gen-2 units. Theoretically. But I can't think of anything that would negate the hypothetical.

How are you figuring this?

It should follow the highest latency. SATA follows this (slowest drive sets the throughput on drives), as does memory that isn't matched. I.e. CAS 7 & CAS 9 would operate at CAS 9.

 

[Tesselator]

I think memory would be a different thing and not comparable to SATA devices on an SATA bus.

Doesn't the stripe distribution alternate between devices at each drive's maximum rate and lowest latency (all other things being equal)? If so (and I believe that is correct) it would be one chunk at gen-1 rates, and then one chunk at gen-2 rates. If we assign numerical values to the rates (say 1 and 2 respectively) it's pretty easy to surmise the average (say, 1.5) rate.

Why would the latency of one device affect the latency of the other just because of data formatting? For SATA devices latency is a constantly unknown factor.




.

 

[aibo]

Thanks for the answers
As long as it would be no slower than using two X25-M gen 1's I'll go for it. Or if the prices on the gen 1's goes even lower than the gen 2's, I may go that route too.

I was just worried that without identical drives in RAID 0, they might have to wait for each other and thus be slower than if I had matched two old X25-Ms.

 

[VirtualRain]

I think memory would be a different thing and not comparable to SATA devices on an SATA bus.

Doesn't the stripe distribution alternate between devices at each drive's maximum rate and lowest latency (all other things being equal)? If so (and I believe that is correct) it would be one chunk at gen-1 rates, and then one chunk at gen-2 rates. If we assign numerical values to the rates (say 1 and 2 respectively) it's pretty easy to surmise the average (say, 1.5) rate.

Why would the latency of one device affect the latency of the other just because of data formatting? For SATA devices latency is a constantly unknown factor.

I agree. Each drive delivers data or commits it to disk as fast as it can. There's no penalty.