Just because you have two drives, it doesn't magically change the MFT of the drives to be half. The MFT of a single unit does not change just because you pair it with another unit.
That's like saying if you have tires on your car with a wear out of 50K miles and because you have 4 tires, that the wear out is now 12,500 miles. It doesn't work that way.
Statistically, the chances of drive (or flat tires) failure is increased, but in real life, the MFT is still 10 years and the tire wear out is still 50K miles.
You're correct in a way.
Pairing drives in a RAID-0 does not increase the chance of each hard drive failing. If the MTBF of each HD is 5 years, then the MTBF of each hard drive is still 5 years when paired in a RAID-0.
BUT ...
The critical thing about a RAID-0 is that data is spread across both hard drives with no redundancy, and therefore the data is now
dependent on
both hard drives operating. So, while the risk of hard drive failure (for each individual drive) hasn't changed, the risk of data loss has gone up.
Another example: Coin flips.
If I flip a coin, there is a 50% chance that the result is heads and a 50% chance that it's tails. This is true of each individual coin flip, no matter what.
If I have already flipped a coin 9 times and got 9 tails results in a row, what is the chance that I will get tails on the 10th coin flip? Still 50%, because that coin flip is independent of the rest. The fact that previous flips resulted in an improbable combination doesn't influence the next coin flip.
But if I ask a different question, that is what are the chances that I flip a coin 10 times and get 10 tails in a row? The chance is 0.097% (1 out of 1024). This is because by asking the question this way, the coin flips are
dependent on each other to create my desired situation. ALL of the flips have to be tails, in a row.
Let's apply the math to the RAID-0:
Typical hard drive has an AAFR (average annual failure rate) of 1.5%. This means that 1.5% of the hard drives of that model will fail in a 1 year operating time frame. Let's assume I use 2 of these drives to create a RAID-0 and operate it for 1 year. What is the chance of data loss due to hard drive failure?
There are 4 situations that can happen in that year:
1. No drives fail.
2. Drive A fails.
3. Drive B fails.
4. Both drives fail.
Let's compute the chances of each situation.
Situation 2 and 3 are easy. The chance of each of those situations happening is 1.5%. 1 drive failure = the manufacturer's stated AAFR.
Situation 4 is extremely unlikely, but possible. The probability of failure is 1.5% * 1.5% = 0.0225 %.
For situation 1, we simply need to subtract all the other situations from 100%:
100% - 1.5% - 1.5% - 0.0225% = 96.9775 %.
So, let's summarize:
[code:1:6db61691d8]
Situation Data Loss? Probability
========= ========== ===========
No drives fail No 96.9775 %
Drive A fails Yes 1.5000 %
Drive B fails Yes 1.5000 %
Both drives fail Yes 0.0225 %
Total probability of data loss: 3.0225 %
[/code:1:6db61691d8]
As you can see, even though each hard drives AAFR (or MTBF) hasn't changed, the probability of data loss in the RAID-0 (3.0225%) is approximately double that of the probability of data loss on a single drive (1.5%). This is because the data is dependent on
both drives being operational.