Originally Posted by assassin
That's basically the same thing I said. I never said that it didn't apply to ANYONE, did I?
No, you asked it the argument was valid
, not if average people should worry about it.
Originally Posted by Mfusick
The same logic could be applied in opposite.
If you purchase all your drives in one batch you could get a good batch and have good luck with all of them.
You can minimize your chances if getting a bad batch by not sampling multiple batches since most batches are very good you have high odds this would work to your benefit.
But you're missing the point, in a redundant array, the loss of a single drive is "no big deal" the array and data are designed to survive that, however loss of more than one drive (or more than your number of parity drives) is catastrophic.
Lets use some completely contrived numbers. Lets say you need ten hard drives and there are ten available batches, 9 of them have 0% chance of failing while one of the batches has a 100% chance of failing.
If you buy one drive from every batch, you have a 100% chance of getting a bad drive, but the array is designed to survive that so since the remaining drives have a 0% chance of failure, your data is 100% guaranteed to be safe, despite getting a failed drive for sure.
If you buy all ten drives from the same batch (chosen at random) you have a 1 in 10 chance of getting drives from the bad batch, a 10% chance. Since the array can't survive more than one drive failing and you have a 10% chance of getting more than one bad drive, you have a 10% chance of losing all your data.
So while buying all from one batch gives you a 10% chance of getting a bad drive (actually multiple bad drives) vs buying from each, which gives you a 100% chance of getting a bad drive, which does sound better, 10% chance of a bad drive seems better odds than 100% you have to consider the use in an array.
Reality is your choice is a 0% chance of losing data by buying from diverse batches vs 10% chance of data loss by buying from a single batch.
This is the argument, and this is why the argument is valid, and not hogwash.
Now reality is the odds are more like 1% or whatever, and the differences may be like 1% for a good batch and 1.01% for a bad batch, and the odds of two drives failing at the same time are like 1% * 1% so, what like 0.01% or something.
I think all these kinds of arguments are just about people being afraid.
People are always afraid of uncertainty so they will act in ways that help them accept it or make them feel better about things they really can't control
Fear is a good thing, it makes you stop and think. People should be afraid of data loss, and should think about it. If you're not afraid of data loss then you've got no reason to back up important data, which is a really bad idea. People should fear data loss and think about it, and decide what data is important and how they want to protect it.
Like I said before, I don't think buying in separate batches is really worth the trouble, but if someone wants to do it, I say go for it. There's solid logic behind the idea and there's nothing wrong with doing it, it's just a little more effort up front.