You are certifiably insane.
The MTTF of my disks is ~4 years. End of story., On the data I have given you.
-- New Socialism consists essentially in being seen to have your heart in the right place whilst your head is in the clouds and your hand is in someone else's pocket.
NO. We have n devices and of those a certain percentage fails after 1 year. THAT is what the MTTF expresses. Not how long it takes for the average unit to fail.
The difference is the measurement period. You are measuring how long it takes for a unit to fail, what the MTxF people measure is what percentage of the units fail within some fixed interval. That is fundamentally different, especially when the interval is short relative to the outcome of the calculation.
It's not uptime that bothers me, it's having a second disc fail before the first is replaced and the rebuild is finished that worries me.
-- Steve O'Hara-Smith | Directable Mirror Arrays C:>WIN | A better way to focus the sun The computer obeys and wins. | licences available see You lose and Bill collects. | http://www.sohara.org/
indeed & if they were all originally installed at the same time the chances of that happening are significant (I have seen it on more than 1 occasion)
-- ISO applications: A solution in search of a problem!
No, it isn't. End of.
Not how long it takes for the average
-- New Socialism consists essentially in being seen to have your heart in the right place whilst your head is in the clouds and your hand is in someone else's pocket.
how likely is that?
-- New Socialism consists essentially in being seen to have your heart in the right place whilst your head is in the clouds and your hand is in someone else's pocket.
Only if they are in the same chassis and got the same voltage spike ;-)
I have NEVER had that happen to me. And we are talking hundreds of machines
=
-- New Socialism consists essentially in being seen to have your heart in the right place whilst your head is in the clouds and your hand is in someone else's pocket.
BTW what is the MTTF of an average asshole's brains ?
JM
The rebuild can stress all the drives more than is sensible at their age.
...but there's a simple solution to that - don't use disks from the same batch, model, or make as your prime disk for your backup disk. Similarly, if you use more than one backup generation, make the backup generations are also different, i.e. do exactly as is recommended for RAID mirroring.
-- martin@ | Martin Gregorie gregorie. | Essex, UK org |
Yes it is. Rob is right; you are wrong.
You appear to be unable to grasp the difference between MTTF and lifetime. There is no direct relationship between them. The only connection is that the MTTF applies during the lifetime. The lifetime is a rating, determined by the manufacturer.
I know that the point of failure of a disc drive is the end of the life of that particular specimen, but it's not the same thing as the manufacturer's rated lifetime.
To be practical: if you have a large enough set of drives (which means large enough to be a statistically representative sample) and you run them, within their ratings, until the end of the rated life as specified by the manufacturer, you would expect that most of them are still operating correctly. In the example I quoted earlier in this thread (5 year lifetime, 1 million hours MTTF), you would expect over 95% of them to be still good after 5 years.
Dave
Rob's logic is correct.
Dave
He is making no such assumption.
The M in MTTF is for Mean, i.e. the average.
But that isn't how MTTF is calculated.
Dave
If you have the replacement drive to hand, start the rebuild immediately and are not using drives from the same batch in the array, and the dead drive is an early failure then probably not very likely. However if you have to wait two weeks for an RMA replacement to arrive (which has happened to me twice) the odds go up sharply, if you built the array using drives from the same batch the odds go up sharply and if all the drives are getting near their expected life then the extra stress of running the rebuild can kill one. Much depends on how many drives you have in the FEC list too - if you have RAID5 with 2+1 then the odds are rather better than if you have RAID5 with 50+1 - both arrays have single drive redundancy but one requires 2 out of 3 drives to fail before you lose data the other 2 out of 51.
-- Steve O'Hara-Smith | Directable Mirror Arrays C:>WIN | A better way to focus the sun The computer obeys and wins. | licences available see You lose and Bill collects. | http://www.sohara.org/
What you think MTTF is, is not what MTTF actually is,.
Go back and read up.
-- New Socialism consists essentially in being seen to have your heart in the right place whilst your head is in the clouds and your hand is in someone else's pocket.
It isn't calculated the way you think it is, else it would be a useless metric.
-- New Socialism consists essentially in being seen to have your heart in the right place whilst your head is in the clouds and your hand is in someone else's pocket.
Of course only a real fool would build a 50+1 RAID5 array, especially when not having one or more hot spares.
However, in practice even with notoriously bad drives in servers at work, it has never happened that a second drive in the array (we usually use RAID1, sometimes RAID5 but only with max 5 drives) fails within months of the first one failing under use. And of course we get a replacement drive within 4 hours of the support call.
It could happen, and it would be really best when different makes of drives are used, but even manufacturers that source their drives from different manufacturers usually don't put different ones in any particular server they deliver. So you would have to arrange for that yourself.
In fact I worry more when the power fails. The chance that a drive fails is significantly more when you cycle the power (and it won't come back up). That could happen to two drives in the array. Never happened to us.
Not all of it. He fails to take into account that the true probability of a failure is not constant over the useful life of a device. He has made statements that you can figure the probability of a failure as if it were a constant and it isn't.
-- Rick
I have made exactly one post in this thread in which I asked a straightforward question that you did not reply to. Let me repeat it. What is the difference between the MTTF of a disk drive and the MTTF of an elephant? Any answer, whether it is the formula used for the calculation or why I should, or should not, use the number in choosing a pet or a disk drive would be appreciated.
I believe that if I base my decision on the MTTF of my pet I would be better off with a pet elephant than with a pet fruit fly. I know that a fruit fly might outlive the actual elephant I chose but, nevertheless, I believe my choice is rational. Am I mislead? Why is a similar choice based on the MTTF of disk drives not equally rational?
Re: Re: choice of external USB drive for mount at boot By: rickman to Andrew Smallshaw on Mon Jun 15 2015 12:39 am
ri> Why do you say higher revs? Maybe I'm not up to date, but the fast ri> spinning drives were always 3.5" because they dissipated a lot more heat ri> and the size helps to get rid of it better. Has that changed?
I haven't seen a 3.5" drive in a Dell desktop for the past year or so -- first they came with Seagate Barracuda XT Hybrid SATAs, now the same 2.5" SATA SSDs they ship in their laptops.
Mind you, I'm an enterprise customer -- YMMV.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.