Utility to burn in new hard drive?

[kony]

yep. Don't expect to keep your data with the test that would
actually produce results.

Excellent and true point.

Not necessarily, having one company use an oven to elevate
temps on *certain* parts doesn't being to make it a standard
for burn-in on everything. The device merely needs be
tested at the bounds of the operating spec if not beyond it.

 

[Joe S]

While that seems right initially, seldom do I hear of a
drive arriving DOA or dying immediately (within an hour),
usually it's within the first 9 months to a year if the
failure is premature.

I have got to sa that there is a lot of truth in that and I hear
about the same sort of thing.

 

[Arno Wagner]

A few drives did have a jumper on the drive that would get it
to exercise continually. Not that common at all, and I havent
noticed any of the current drives with that capability now.

I am not disputing that. Modern drives often allow automatic
execution of a short SMART self-test every few hours, to be
set by software. But they cannot ''submit'' diagnostics to
the computer. They can only store them and the computer has
to actively ask in order to get any information.

Arno

 

[Osiris]

With decent scripting, not at all. Don't know whether the "market
leader" has finally added a shell that allows to do that to its
product....

don't think so...
1: they want to sell you a second drive. For that you must be a
satisfied consumer. they get there by producing a drive that is
(subjectively) good *enough*.
2: they want to ease you past the warranty date, by not letting you
rocking the boat.

I go for 1: any for-profit organization is commited to maximizing
longterm owner value, by selling goods or services.
And a good measuring stick is consumer opinion which translates (not
1:1) in owner value.

 

[Joe S]

I thought one of the differences between a Maxtor DiamondMax and

Nope, no one does that anymore.


It makes more sense to increase the backup frequency for the first
couple of months or so.

I still like the eight or nine drive raid configuration
where even if a drive does take a dump, NO data gets lost,
and the replaced drive gets the data image rebuilt automatically.

THAT is the most reliable method, and if this guy IS running a
business, and is setting up a server that he wants to be
completely reliable, this IS the best option.

SCSI and IDE raid controllers that perform that level of raid
array are not cheap though.

I am in the process of making a eight or nine drive raid level
3 array with laptop drives. I may even set it up for hot swap
for repairs on the fly. [snipped]

I am the OP and I wish I could afford something like this 8 or 9 disk
RAID! My system is at home and is used for home purposes. The data
I store on my system is about 500GB of audio/video which I have
created myself and which I do not want to lose.

I do my backups about once a week. In between backups I could get a
failure.

To avoid the failure of one spindle could I use a 2 disk RAID.
Presunably the protection this gives would be similar to backing up
one of the two disks (on a continuous basis). Is this correct?

 

[JohnH]

I am not disputing that.

You did with that 'nonsense'

Modern drives often allow automatic execution of a short
SMART self-test every few hours, to be set by software.
But they cannot ''submit'' diagnostics to the computer.

He wasnt using the word submit like that, just
saying that diagnostics can be run. He was right.

They can only store them and the computer has
to actively ask in order to get any information.

That is what he said in different words.

 

[Rod Speed]

I thought one of the differences between a Maxtor DiamondMax and
a MaXLine was that the MaXLine had been soak tested for longer?

Nope, no one does that anymore.

In that case, testing a new drive might be worthwhile?

It makes more sense to increase the backup frequency for the first
couple of months or so.

I still like the eight or nine drive raid configuration
where even if a drive does take a dump, NO data gets lost,
and the replaced drive gets the data image rebuilt automatically.

THAT is the most reliable method, and if this guy IS running a
business, and is setting up a server that he wants to be
completely reliable, this IS the best option.

SCSI and IDE raid controllers that perform that level of raid
array are not cheap though.

I am in the process of making a eight or nine drive raid level
3 array with laptop drives. I may even set it up for hot swap
for repairs on the fly. [snipped]

I am the OP and I wish I could afford something like this 8 or 9 disk
RAID! My system is at home and is used for home purposes. The data
I store on my system is about 500GB of audio/video which I have
created myself and which I do not want to lose.

I do my backups about once a week. In
between backups I could get a failure.

'burning in' the drive isnt really going to help in that regard.

Like I said, it would make more sense to increase the
backup frequency in the first couple of months instead.

To avoid the failure of one spindle could I use a 2 disk RAID.
Presunably the protection this gives would be similar to backing
up one of the two disks (on a continuous basis). Is this correct?

No, that wont protect against user stupidity, deleting a file in error.
Or the house burning down etc either. Or the system getting stolen.

It would help with single drive failure, but that isnt that
likely with modern drives and that is what backup is for.

The only real advantage with RAID is that you do have
another copy all the time and you can recover more quickly
if a single drive fails. Not if the raid controller fails tho.

How are you doing your backups ?

 

[Phat Bytestard]

Doesnt protect against some failures.

Oh boy... catastrophic power line surges or power supply
failures... not much else since a multiple drive failure is highly
unlikely and extremely rare.

Nope, mirroring over multiple machines is.


Nope, mirroring over multiple machines is better.

Mirroring to alternate RAID 3 machines is better still.
Providing an even higher chance that a power line surge type failure
as described above will not fail through multiple machines.

And if it dies, you're still fucked.

Nope. There would be nothing stoping one from placing a new drive
controller in the machine. A good IT admin would have a backup piece
of hardware on hand. Many "modern" IT admin twits have gotten far too
casual and trusting of the gear.

Insanely over priced compared with mirroring over multiple machines.

Not at all. The arrays can be placed on multiple machines and be
utilized as mirrors. This provides multiple levels of raid
functionality.

Nope, its still the bleeding edge capacity wise.


Could actually, mirroring over multiple machines is even more
perfect. That way whatever dies the system carrys on regardless.

Not if both die. Ooops! The same problems you claim I would suffer
could easily be suffered across multiple machines as well.

 

[Phat Bytestard]

There are no 'factory burn in labs' with mass market commodity drives.

Sorry, but they all undergo live testing in a hot lab.

You in spades.

Go look in a mirror. It seems to be your favorite word.

 

[Phat Bytestard]

No obvious place to **** it unless you have a
needle dick and stick it in the molex connector.

That is the most stupid post you probably have ever made.

 

[Phat Bytestard]

Not necessarily, having one company use an oven to elevate
temps on *certain* parts doesn't being to make it a standard
for burn-in on everything.

A "burn-in test" is performed on complete products, not individual
components thereof.

Component testing and qualification takes place at the component
makers locale.

The device merely needs be
tested at the bounds of the operating spec if not beyond it.

THAT is NOT a "burn-in" test.

 

[Phat Bytestard]

I am not disputing that. Modern drives often allow automatic
execution of a short SMART self-test every few hours, to be
set by software. But they cannot ''submit'' diagnostics to
the computer. They can only store them and the computer has
to actively ask in order to get any information.

Yes, but the computer user and OS does not have to do ANYTHING for
the "repair" to take place.

Drives map out bad sectors. OSes do not need to intervene at all.

 

[Phat Bytestard]

I thought one of the differences between a Maxtor DiamondMax and
a MaXLine was that the MaXLine had been soak tested for longer?

Nope, no one does that anymore.

In that case, testing a new drive might be worthwhile?

It makes more sense to increase the backup frequency for the first
couple of months or so.

I still like the eight or nine drive raid configuration
where even if a drive does take a dump, NO data gets lost,
and the replaced drive gets the data image rebuilt automatically.

THAT is the most reliable method, and if this guy IS running a
business, and is setting up a server that he wants to be
completely reliable, this IS the best option.

SCSI and IDE raid controllers that perform that level of raid
array are not cheap though.

I am in the process of making a eight or nine drive raid level
3 array with laptop drives. I may even set it up for hot swap
for repairs on the fly. [snipped]

I am the OP and I wish I could afford something like this 8 or 9 disk
RAID! My system is at home and is used for home purposes. The data
I store on my system is about 500GB of audio/video which I have
created myself and which I do not want to lose.

Back up... a lot. to optical disc, off machine HDs or other
device.

BTW, a bit striped drive also has a higher throughput.

I do my backups about once a week. In between backups I could get a
failure.

Good practice.

To avoid the failure of one spindle could I use a 2 disk RAID.
Presunably the protection this gives would be similar to backing up
one of the two disks (on a continuous basis). Is this correct?

That is a mirrored array.

 

[Phat Bytestard]

The only real advantage with RAID is that you do have
another copy all the time and you can recover more quickly
if a single drive fails. Not if the raid controller fails tho.

Bullshit. The bit striped array I described carries with it a much
higher throughput. Also, a controller failure is even less likely
then a drive failure since it is not a mechanical device. Also,
replacing it would bring the system back up with no loss of data,
which is the goal. Cost is not a factor.

 

[Rod Speed]

Yes, but the computer user and OS does not have
to do ANYTHING for the "repair" to take place.

Wrong with sectors that cant be read. The hard drive wont
replace those until a write is attempted to that sector.

Drives map out bad sectors. OSes do not need to intervene at all.

Its more complicated than that, most obviously with sectors that cant be read.

 

[Rod Speed]

Oh boy... catastrophic power line surges or power
supply failures... not much else since a multiple drive
failure is highly unlikely and extremely rare.

Failure of the RAID hardware and the PC its in isnt extremely rare at all.

Mirroring to alternate RAID 3 machines is better still.

Nope, pointless if its mirrored over multiple machines.

Providing an even higher chance that a power line surge type
failure as described above will not fail through multiple machines.

Its much better to ensure that power line surges cant fry anything instead.

Yep.

There would be nothing stoping one from
placing a new drive controller in the machine.

Assuming its even obtainable anymore. And that
it didnt **** the array in the process of dying too.

A good IT admin would have a backup piece of hardware on hand.

Dont need that if its mirrored over multiple machines.

ALL you need to do even if the entire machine fails is to
replace it with another machine and new machines will always
be available since there isnt anything unique about them.

Many "modern" IT admin twits have gotten
far too casual and trusting of the gear.

You in spades, 'trusting' the raid hardware and the machine its in.

Not at all.

Corse it is.

The arrays can be placed on multiple machines and be utilized as mirrors.

No point with multiple machines.

This provides multiple levels of raid functionality.

Completely pointless with mirrored multiple machines.

Not if both die.

Taint gunna happen if you've got enough of a clue to ensure that
no mains surge can do that, and thats completely trivial to ensure.

Ooops! The same problems you claim I would suffer
could easily be suffered across multiple machines as well.

Nope.

 

[Rod Speed]

Bullshit.

We'll see...

The bit striped array I described carries with it a much higher throughput.

Bullshit it does. In spades when you are stupid enough to use laptop drives.

Also, a controller failure is even less likely then
a drive failure since it is not a mechanical device.

Still a single point of failure. Multiple machines dont have that.

Also, replacing it would bring the system back up with no loss of data,

Depends on how the raid hardware dies.

which is the goal.

Pity that aint guaranteed with raid.

Cost is not a factor.

The multiple machines leaves it for dead.

 

[Rod Speed]

Liar.

but they all undergo live testing in a hot lab.

That aint a 'factory burn in lab'

Go look in a mirror.

No thanks, they keep breaking when I do that.

It seems to be your favorite word.

No it isnt. I didnt even use it in that post.

 

[Rod Speed]

A "burn-in test" is performed on complete
products, not individual components thereof.

Wrong, as always. And he said parts, not components anyway.

Component testing and qualification takes
place at the component makers locale.

Pity about hard drives.

 

[Arno Wagner]

You did with that 'nonsense'

He wasnt using the word submit like that, just
saying that diagnostics can be run. He was right.

That is what he said in different words.

Well, ok then. Might have been an overreaction on my part.

Arno