Realiability: (1) 4TB or (2) 2TB HDD?

[MBony]

After reading reviews on Seagate's 4TB drive and seeing Backblaze's report (http://gigaom.com/2014/01/21/backblaze-on-cheap-hard-drives-buy-hitachi-if-you-can-afford-em/) on hdd realiability I was wondering which I should go with if pricing is relatively the same.

 

[mfenn]

Are you talking about JBOD or RAID 1? If JBOD, the single drive will be more reliable. If RAID 1 the 2TB drives will be more reliable.

 

[MBony]

Are you talking about JBOD or RAID 1? If JBOD, the single drive will be more reliable. If RAID 1 the 2TB drives will be more reliable.

Does how they are configured matter to the integrity of the disk? I'm wondering if smaller drives will last longer and create fewer bad sectors than larger drives when put under the same stresses.

 

[Steltek]

Does how they are configured matter to the integrity of the disk? I'm wondering if smaller drives will last longer and create fewer bad sectors than larger drives when put under the same stresses.

The MTBF for modern hard drives is quite long, so it isn't usually a problem. Installing your hard drives in a well ventilated case to reduce heat is typically the best thing you can do to increase longevity. Reliability is a fickle thing - Hitachi might be the most reliable in that report, but in a few months it might be WDC or Seagate.

You should instead worry about protecting yourself from hardware failure by creating an automated backup regimen. The automation part is really important because, human nature being what it is, backups that aren't automated usually don't get done until it is too late.

 

[ViRGE]

Does how they are configured matter to the integrity of the disk? I'm wondering if smaller drives will last longer and create fewer bad sectors than larger drives when put under the same stresses.

RAID 1 is mirroring. So if one drive failed you'd still have everything on the second drive.

Otherwise if you're talking about placing data on both drives, then the 4TB drive is the safer bet. The odds of a drive failure go up with the number of drives since each drive is an independent event.

 

[Cerb]

Does how they are configured matter to the integrity of the disk? I'm wondering if smaller drives will last longer and create fewer bad sectors than larger drives when put under the same stresses.

Maybe, maybe not. By the time enough data can be accrued, assuming anyone is willing to divulge it (Google was not, FI), the drive will be on its way out, with new replacements coming in.

Two drives is generally less reliable just due to having more parts to fail. When large capacities are brand new on the market, they are typically still working out the kinks in manufacture and QC, and they are less reliable. But, 4TBs are now pretty mature.

 

[Charlie98]

If you have 4TB of data to store, the correct answer is 2x 4TB drives...

 

[BrightCandle]

If the point is having 4TB of storage then a single drive is going to be more reliable than 2x2TB drives.

<look away now if you don't like maths>
Lets say each drive has a 10% chance of failure then the two scenarios are:

Scenario 1
Single 4TB drive
10% failure, 90% survival

Scenario 2
2x 2TB drive for 4TB of space.
Chance the array survives = 0.9*0.9 = 81% survival
100-81 = 19% chance of failure

Scenario 3
2x 2TB drive in raid 1 (assuming you don't replace a damaged drive in the year) = 2TB of space

Survive = Both Survive(0.9 * 0.9) + (D2 fails)(0.9*0.1) + (D1 fails)(0.1*0.9) = 99%
Failure = 1%

Of course if you replace the drive when one of them fails then even if that takes you all year its 1%, if you do it right away its a lot less than that.

So yeah Raid 1 decreases the chance of a loss of data considerably.

 

[Essence_of_War]

After reading reviews on Seagate's 4TB drive and seeing Backblaze's report (http://gigaom.com/2014/01/21/backbla...can-afford-em/) on hdd realiability I was wondering which I should go with if pricing is relatively the same.

I wouldn't put too much stock in that report for evaluating my purchase of modern 4 TB, 1 TB platter drives. For example, some of lehtv's thoughts on this.

The drive with the highest fail rate, ST31500341AS (25.4%), is a Seagate 7200.11 drive from 2008. A six years old model using three 500GB platters. This makes me doubt there are any useful conclusions to be drawn from this drive's survivability when compared to newer 1TB per platter drives. It is clearly an outlier which you can't use to judge Seagate's reliability in general, especially that of their current drives. The second highest failure rate was on ST31500541AS (9.8%), much improved over the -341AS. But even this one is a five year old model, not exactly relevant any more. For example, ST4000DM000 is a new drive with 3.8% failure rate, similar to WD's numbers.

I don't think these results should be taken as "Seagate sucks, Hitachi rocks", one has to look at each model separately instead of blindly trusting one brand over another, and also weigh the cost of the drive and the length of its warranty against the likelihood of it failing.

Emphasis mine.

 

[mfenn]

If the point is having 4TB of storage then a single drive is going to be more reliable than 2x2TB drives.

<look away now if you don't like maths>
Lets say each drive has a 10% chance of failure then the two scenarios are:

Scenario 1
Single 4TB drive
10% failure, 90% survival

Scenario 2
2x 2TB drive for 4TB of space.
Chance the array survives = 0.9*0.9 = 81% survival
100-81 = 19% chance of failure

Scenario 3
2x 2TB drive in raid 1 (assuming you don't replace a damaged drive in the year) = 2TB of space

Survive = Both Survive(0.9 * 0.9) + (D2 fails)(0.9*0.1) + (D1 fails)(0.1*0.9) = 99%
Failure = 1%

Of course if you replace the drive when one of them fails then even if that takes you all year its 1%, if you do it right away its a lot less than that.

So yeah Raid 1 decreases the chance of a loss of data considerably.

:thumbsup: Thank you for doing the math that I was too lazy to work out.