WD caviar 1TB MTBF, any clues?

[aphex]

What kind of MTBF figures can I expect on my HDD in my B2? (It's a 1TB WD Caviar one)?

My b2 has been running for few years now, and smartmontools tells me then "Power_On_Hours" is 15170 (2 years), Power_Cycle_Count 51 everything else under 50.

Given how vital my bubba is to my life (yeah, it really is) I'm starting to think about "what if", and when "what if " might happen.

I read somewhere online that MTBF for these kinds of drives are about 30000 hours? I do realize that this depends on temperature, spin-up count etc etc. But let's assume temp and spin-ups has been kept well within acceptable levels.

Basically I am asking the Excito guys (and any other expert here) what kind of figures I can expect really. Have anybody here had their bubba HDD crash, and under what circumstances?

If we get a scientific answer to this, that might be something for the wiki.

[aphex]

This is what WD is saying btw...

http://wdc.custhelp.com/app/answers/det ... WksxSGs%3D

And here some more good info about Google's experiences

http://storagemojo.com/2007/02/19/googl ... xperience/

[Nrde]

What kind of MTBF figures can I expect on my HDD in my B2? (It's a 1TB WD Caviar one)?

Short answer: nobody knows.

Long answer: if it hasn't broken already it should last a long time still.

(yeah, not that helpful answer)

[Binkem]

Just make sure you have a backup, because failure may (and will) strike at any moment.

[Ubi]

Not to piss in anyone's cheerios but...
MTBF is completely irrelevant if you only have a single drive, as you know nothing of the spread around the mean. The number is meant for data centers.

[Ubi]

Here's some reading to backup my provocation
http://www.eweek.com/c/a/Data-Storage/H ... -or-Farce/
http://db.usenix.org/events/fast07/tech ... index.html

[Gordon]

Not to piss in anyone's cheerios but...
MTBF is completely irrelevant if you only have a single drive, as you know nothing of the spread around the mean. The number is meant for data centers.

Seconded...

There may be something to consider though if you've enabled mirroring. It's kind of a stupid thing, but I have it seen go bad once or twice. The thing is that you're likely to go into the shop and pick up two identical drives from the same manufacturer. Since you're buying them as a pair, chances are big that they came from the same wholesale box and where manufactured the same day on the same production line and supervised by the same person. If a problem occurred that day and this causes one of the drives to fail after a certain time, there is a serious hazard that the second drive may also fail just minutes after the first one failed.

[Ubi]

Yeah those events are amusing to watch from a distance. WHen it was my own raid set, not so much

[Gordon]

I don't recall saying amusing and you can trust me that being the one to tell a customer that his backups are bad as well is absolutely no fun at all.

[Ubi]

nope you did not, but I did. And I guess this customer was not your customer when he set up his backup solution. It's kinda interesting that this person used the same drive size for backup as for production. This rather reeks not of backup but of a RAID1 with very high latency...

But all schadenfreude aside, it is not easy these days to purchase a server with disks from distinctly different batches. The only thing really to guard against is is to buy your spare drive from another vendor at a different time point. Purchasing will hate you for it though...

[Gordon]

It is actually multiple customers and I'm not really maintaining their network but providing an application that counts as mission-critical in most cases. All of these customers are aware of the fact that they need to protect their data and they all use RAID configurations. What they regularly don't do is make sure their backups are done correctly, and of course this is something they discover when the protection they expected from RAID has failed.

Honestly? I think it is cool that the B3 can do RAID1, but I see really no use for it in a home situation. There´s absolutely no telling when a disk might fail and if one did then there´s this little thing that you need to become aware of the fact that one of the disks failed. I´ve actually been to a company once where they didn´t realize that their main system had failed and they´d been working on a standby system for over a week already.

Best advice I can give: make backups. Do it daily if required, but at least once a week. And most importantly: make sure your backups are consistent.