fast single platter drive

[ubersonic]

Rubbish. The platters, arms and heads are are linked together on a multi platter drive. Single platter drives would have a lower mtbf if what you were saying is true.

Please don't call other peoples information rubbish just because you don't understand it, more heads = more heads that can fail, its not that difficult a concept. And if what i'm saying is true (and it is) the single platter drives would have a higher mtbf not lower because that's how mtbf works.

"the greater number of components in a system, the more frequent a hardware component will experience failure, leading to a reduction in the system's hardware MTBF"

 

[CaptainCrash]

Please don't call other peoples information rubbish just because you don't understand it, more heads = more heads that can fail, its not that difficult a concept. And if what i'm saying is true (and it is) the single platter drives would have a higher mtbf not lower because that's how mtbf works.

Where's your data?

 

[MagicBoy]

Please don't call other peoples information rubbish just because you don't understand it, more heads = more heads that can fail, its not that difficult a concept. And if what i'm saying is true (and it is) the single platter drives would have a higher mtbf not lower because that's how mtbf works.

"the greater number of components in a system, the more frequent a hardware component will experience failure, leading to a reduction in the system's hardware MTBF"

The perils of concentrating on typing on a smartphone and not reading what you stab out on a touchscreen!

I perfectly understand the concept, I just don't agree with you. I'm perfectly well qualified to comment, thanks for asking.

 

[Lakeland]

"the greater number of components in a system, the more frequent a hardware component will experience failure"

True. The chance of failure might be so low due to over engineering that only an error in production would cause failure (unlike a car where parts wear and Will have to be replaced). But anyways, you're right, not taking account of how significant the chance of failure increases, it is absolutely there.

 

[AbsenceJam]

More parts does not mean more actual failures, indeed in some situations it can be the opposite. There is no data to suggest that a single platter drive is any more reliable than one with 4 platters. Manufacturing defect, shock, etc would make both as susceptible.

 

[ubersonic]

Where's your data?

Its called "the laws of physics", seriously what are people not getting about this?

OK here's some pretty analogy:

(1) The older F1 fans on here may remember the 6 wheeled F1 car that competed in the 70's (if your 16 and are amazed to hear of this then wiki the Tyrrell P34), well that had more tyres than the other cars and as a result had higher chances of having a blow out.

(2) The backbone of the US air force fighters is the F16 Falcon, it is a single engine 4th gen fighter that is cheaper to build/maintain that all the other fighters they use (F15/18/22/etc) as it only has one engine and thus the chances of an engine failure are lower than on the twin engine jets (yes I know that a single failure would be catastrophic unlike on the other jets but i'm not citing redundancy here merely the chances of one engine failing)

So you see the more heads you have the higher the chance of head failure, the more platters the higher the chance of platter failure.

it is totally possible right now for a HDD manufacturer to make a 16TB drive (32 platters at 500GB) but they won't do it because it would be huge and hotter than the sun and the failure odds would be immense. Back in the 90's we regularly saw 10 platter drives, today one of these could house 5TB but again the odds of something going wrong would be high.

Single platter versions of the same drives have better reliability odds than their multi plater brothers because the is less to go wrong and the drives run cooler thus reducing the odds of a failure further

 

[gazzamcg]

There's no moving parts with an ssd, but they can still break down.

 

[CaptainCrash]

Single platter versions of the same drives have better reliability odds than their multi plater brothers

Again, can you produce *evidence* to that effect, as opposed to sweeping generalisations, oversimplified theory and assumptions based on dodgy analogies?

edit: it would also be helpful if you would be more precise about "multi-platter" in this context, presumably we're not actually talking about 32-platter drives.

 

[MagicBoy]

OK here's some pretty analogy:

(1) The older F1 fans on here may remember the 6 wheeled F1 car that competed in the 70's (if your 16 and are amazed to hear of this then wiki the Tyrrell P34), well that had more tyres than the other cars and as a result had higher chances of having a blow out.

I don't recall the P34 being particularly susceptible to punctures. Not exactly a heroic failure either - I seem to remember it did win a race with 1-2 finish. Anyway if one front tyre deflated there's always another to help you get back to the pits. The old redundancy argument again.

<snip>
So you see the more heads you have the higher the chance of head failure, the more platters the higher the chance of platter failure.

Platter failure! Don't think in all my years I've ever seen a hard drive with platter failure.

Head crashes - yes
Stuck heads - yes
Failed motors - yes
Blown controller chips - yes

Never a shattered/failed platter. Would require some sort of epic fail to damage a solid aluminium alloy disc.

 

[ubersonic]

Again, can you produce *evidence* to that effect, as opposed to sweeping generalisations, oversimplified theory and assumptions based on dodgy analogies?

Nope nor can I produce evidence to say that gunshots are dangerous to humans as nobody has done an in depth study into the bloody obvious, however if you really wont believe logic without hard evidence then at lunch time go down to the nearest school and ask to speak to the head of maths, then ask them to explain probability to you.

I don't recall the P34 being particularly susceptible to punctures. Not exactly a heroic failure either - I seem to remember it did win a race with 1-2 finish. Anyway if one front tyre deflated there's always another to help you get back to the pits. The old redundancy argument again.

The P34 was a very good car, ahead of its time, in fact one privately owned car is still winning races today. But like I said I wasn't talking about redundancy (and unless were talking about a HDD with internal raid1 it isn't applicable) my point with that example was the more tyres you put on a car the more the chance of one failing increases.

Platter failure! Don't think in all my years I've ever seen a hard drive with platter failure.

Sorry I was referring to bad sectors, all drives have some of them due to manufacturing defects, and wear and tear of platter surface over the years will cause more as will head crash and tracking errors.

 

[AbsenceJam]

I'd suggest you ask an English teacher to explain what "baseless" and "assumption" mean.

 

[ubersonic]

I'd suggest you ask an English teacher to explain what "baseless" and "assumption" mean.

Don't worry I already know, but as my statements were based on fact I fail to see the relevance of baseless to the discussion, however you are correct that I made an assumption and with retrospect you are right it was foolish of me to assume that just because the laws of probability have not altered since the dawn of time that they would be the same now as when I left university.

** EDIT**

Joking around aside this argument has really been going on too long now, Its obvious ive failed to explain it to you guys well enough so ill just use some quotes instead

There's even a potential reliability benefit to a drive with fewer platters. Head crashes are the most catastrophic of common hard drive failures. With fewer platters presenting themselves as potential candidates for warping or targets for a wandering drive head, the F1 has a statistically lower chance of failure than drives with four or five platters.

More platters + more heads equals higher chance of failure.

 

[ecksmen]

Don't worry I already know, but as my statements were based on fact I fail to see the relevance of baseless to the discussion, however you are correct that I made an assumption and with retrospect you are right it was foolish of me to assume that just because the laws of probability have not altered since the dawn of time that they would be the same now as when I left university.

** EDIT**

Joking around aside this argument has really been going on too long now, Its obvious ive failed to explain it to you guys well enough so ill just use some quotes instead

For the record, I don't think your wrong. However, the bigger picture is really the key here. You seem to only care about reliability, therefore you shouldn't be running with a single copy of the file, you should have copies off site etc... You have to assume it will fail, regardless. Until you have removed all possibilities, it is possible and you should plan that it will happen.

 

[CaptainCrash]

Don't worry I already know, but as my statements were based on fact I fail to see the relevance of baseless to the discussion, however you are correct that I made an assumption and with retrospect you are right it was foolish of me to assume that just because the laws of probability have not altered since the dawn of time that they would be the same now as when I left university.

** EDIT**

Joking around aside this argument has really been going on too long now, Its obvious ive failed to explain it to you guys well enough so ill just use some quotes instead

What's really been going on for too long is your gratuitous rudeness and belligerence.

No, the laws of probability haven't altered, and neither has the scientific method (google it). You've yet to provide any evidence that in practice, a single-platter drive has a lower failure rate than, say, a two or three platter equivalent.

If it's so "bloody obvious", then there should be plenty of corroborative data to that effect, surely?