I asked this before but no one seemed to know, but anyway; Is the 500gb version as fast as the 1tb version? I asked because its about half the price meaning i could get 2 for raid As i do not need 1tb for my C drive
In the raid that you'd be going with (RAID 0) for performance, I'd get the 2 500GB drives in raid. There isn't a huge difference in the speed of the two drives. There will be more space of the faster (outer part) of the drive on the 1TB, but I'd doubt it would come close to raid 0.
Remember that RAID significantly increases the likelihood of failure. If a hard drive has a likelihood of failure each year of 7.5%, the same setup but with two drives will have a chance of failure of over 14%
If 1 drive fails you're screwed. Basically you're using 3 drives as one, so you have 3 drives that could possibly fail, instead of just one.
Well it doesn't really increase the "chance of failure", but the chances are that compared to 1 drive, having three there is a three times chance that one of them will develop a fault. And the result of that being losing all data on the array
There's more potential points of failure. If you have one drive, there's one motor, one head, one controller board etc. If you have three drives, you have three times as many parts, of which only one breaking would leave all your data unrecoverable. In your case, assuming that there is a failure rate of 7.5% with hard drives, year on year, there's a 20.8% chance of it failing. If the chance is only 5%, then the failure rate would be 14.26% The probability of the array failing is 1 - the probability of none failing. The probability of none failing is (1-chance of one failing)^number of drives so, probability of array failing is 1 - (0.925)^3 = 20.8
Ah right. Thanks for clarifying that. I back up my programs and os that are stored on C:\ after each program install - well major ones anyway and keep it defragged. Should last me a while yet.
I swear to god i explained that to you in great detail the other day sam... anyway you don't think there is much difference in speed?
No, the raid will be faster, due to how the data is managed, as long as your raid card can handle it. Whatever you buy will be faster than what you've got now. You did explain it, but bakes used maths to prove it.
RAID0 is rarely noticeable outside synthetic benchies - the risk of data loss just isn't worth it imo. Much better off with RAID5/6. jbloggs - For the most part it's simple statistics... If you have one component with a certain failure rate, running two of them together (without any effect on each other's reliability) doubles the chances of failure. That's before you get to the increased chances of data corruption during mirroring.
Raid 0 - Why? You don't need statistics to prove that the MTBF of a Raid0 array is the MTBF of one drive / number of drives. So if the MTBF is 1000 hours with one drive. It'll be 500 hours with two, 250 hours with four etc. So basically a two drive Raid0 drive is "statistically" likely to fail in half the time of a single drive (more or less). It doesn't mean it will, we're just playing with numbers here after all. http://en.wikipedia.org/wiki/Standard_RAID_levels#RAID_0_failure_rate And given that Raid 0 has almost no day to day real world noticable benefits, many will say the risk is not worth it.
I was simply asking where he got the figure of "7.5%" for the likelihood of hard drive failure each year and nothing else... ________ CLASSACTION SETTLEMENTS
http://labs.google.com/papers/disk_failures.pdf I knocked a little bit off Google's numbers, but their drives are likely not in heavy use for most of the time due to the fact that emails and suchlike are accessed periodically by most people so I'd hope that it's fairly representative. Also note that Google's drive tend to be enterprise quality drives, which are built to higher standards and have lower failure rates because of this.
Thanks...yes, I think I remember that Google study from a while ago, when it was released... ________ CLASS ACTION LAWSUIT
you wont really notice much improvement i wouldnt bother with raid 0. The risk factor just aint worth it.
Chance of failure isn't (chance of none failing)^number of drives... The chance one fails is 7.5% per year, that's 7.5*3% per year for 3 HDD's, not 0.925^3, so that's 22.5% per year. You are -with that formula- implying that an array of 500 HDD's has a 0.0000000000000001% chance of failing. //Edit// *Smacks self* That's 1 - chance^number of drives
