[ale] [semi OT] encouraging and discouraging HDD and SSD observations

Ron Frazier (ALE) atllinuxenthinfo at techstarship.com
Mon Nov 4 15:17:25 EST 2013


Hi Brian,

You make some good points about the economics of this industry.  I want to respond to a couple of minor things.

Lots of snipping.  See below.

Sincerely,

Ron



Brian MacLeod <nym.bnm at gmail.com> wrote:

>
>On 11/1/13 8:44 PM, Ron Frazier (ALE) wrote:
>

-snip-

>
>
>Fault #1: your observations are not the end all - be all of how
>everyone else experiences things, but you are making assumptions based
>on your experiences, which, in part, are based on your choices.
>

Obviously.  My experience is my experience.  I would be delighted if my experience is worse than average and if I get other brands / models of drives in the future and they last longer.

-snip-

>
>
>I ride a motorcycle.  Mine is a used unit.  I am a casual weekend
>rider, so it generally gets light usage.  But say one day, I decide
>I'm going to take a trip to Road Atlanta and race it.
>
>Do you really think that bike will survive that?  NO.  Mixing usage
>cases for mechanical equipment is generally one of the WORST things
>you can do.  If it needs to be light, keep it light.  If it needs to
>be intense, never let up, but don't shift gears on equipment like that
>or you will help cause the failure.

I think it's a good idea to periodically run mechanical devices above their average use case.  NOT to redline it.  NOT to abuse it.  NOT to race it, using your analogy.  But to get it a bit above it's typical use case.  So, for a car, I might drive it in 3rd gear rather than 4th for a few miles once in a while, just to get the tach up to 3500 instead of always cruising at 2500, etc.

>
>You adamant adherence to Spinriting the hell out of your drives after
>periods of light usage is part of why you are killing your own drives,
>and then blaming it on the manufacturer.
>
>If you're going to be hard on your drives, ever, keep it that way.
>
>

There are a couple of erroneous assumptions here.

A) The implied assumption is that I do this all the time.  I don't do this process frequently.  Usually only 2-3 times per year.

B) The other assumption is that this tactic is particularly hard on the drive.  It's not.

Spinrite does basically the same thing badblocks does, if you don't count it's statistical data recovery function in the event of read failures.  And if you get a read failure, well, it's already failed.  So what if you try hard to recover the data.

When used for maintenance, it's very similar to badblocks.  In read only mode, it's similar to badblocks read only mode.  In read / write mode, it's similar to badblocks nondestructive read / write mode.

In read only mode, you are simply reading every sector on the disk and assuring it's readability.  This allows the controller to flag any sectors it doesn't like through its diagnostics and potentially reallocate them, before they get so weak that they cannot be read at all.

In read / write mode, it reads the data, inverts it, writes it, reads, inverts, and writes again.  This has the effect of refreshing their magnetic fields if they are weak for ANY reason, including bit rot, weak magnetic coating, whatever.  Again, the controllers diagnostic routines are continually scanning for any surface areas that are excessively unreliable.  

In terms of mechanical stress, this is a very tame procedure.  Let's say the drive has 100,000,000 sectors and 100,000 tracks.  So, the utility just parks the head servo over track 1, does all it's reading and possibly writing for all the sectors in that track, then moves on to track 2, etc.  It continues doing this until it gets to the last track, track 100,000.  This may take days for a large drive.  So, the heads move from one end of the platter to the other over the course of hours or days in 100,000 little steps.  This is an extremely minimal stress compared to defragging the drive, non linear editing, or database access.

I assure you, running Spinrite on the drive, or running badblocks, a couple of times per year doesn't substantially reduce the mechanical reliability of the drive.  It does, however, substantially increase the magnetic reliability of the drive.

IF Spinrite finds an unreadable sector, it WILL fly the heads over to that sector from different positions and at different speeds up to 2000 times to try to recover the data.  HOWEVER, running it for maintenance purposes periodically helps prevent any sectors from ever becoming unreadable for magnetic reasons.  And, thus, the recovery mode is rarely invoked unless the drive is starting to show mechanical failure modes.  In those cases, Spinrite can often recover the data well enough before the drive fails to save it elsewhere.

This process DOES work the electronics and the electronic portion of the read write heads hard.  But, after a burn in period and barring electrical surges and brownouts, etc., the electronics of hard drives are not usually what kill them.

-snip-

>
>> The drive I'm about to replace was a replacement.  It's a refurb
>> that they sent me to take care of my last warranty claim less than
>> a year ago.  The new warranty period only extends to the end of the
>> original period.  I have no problem with that.  But, the refurb
>> drive has been in service less than 1 year.  There is simply no way
>> it should be doing throwing errors at this time, especially since
>> it is supposed to be a replacement for a 5 year drive.
>
>
>It's a refurb.  I expect it to fail.  I plan for it to fail.  Your
>failure to plan for it does not mean the company is at fault.
>
>

Are yo saying you expect refurbs to be inferior to new just because they frequently are?  Or are you saying they should be that way?  If I were going to refurb my product and resell it and put a warranty on it just like it was new, I'd prefer that it was as good as new when it left the refurb shop.

-snip-

>By the way, I now require you to have a 5 year warranty on anything
>you ever sell me, because that's how it should be, and there's no
>wiggle room here.
>
>

If you contracted with me to buy product X and you required a 5 year warranty as part of the contract, then you'd get a 5 year warranty.  I have two choices to accommodate you.  A) I can build the quality into the design such that I don't anticipate a warranty claim but I am prepared to fill one from a minimal number of customers.  B) I can build mediocre quality and essentially buy an insurance policy on your product to cover the cost of replacement if you do make a claim.   I personally would chose option A if i can.  I want the product to be quality.  I want you to be happy.  And, I don't want you to have the hassle of making a warranty claim and I don't want the hassle of servicing one.

If I had to build your product a million at a time, and I couldn't justify building the higher quality product, then I would have no choice but to give you the increased warranty using insurance rather than intrinsic quality.

Either way, I'm going to charge you extra.

In the example I gave in my original post, I, as a consumer, am shopping at Frys for hard drives.  I look at two WD hard drives sitting there on the shelf.  One is WD Green, with a 2 warranty.  The other is the WD Black, with a 5 year warranty.  I, as a consumer, am going to pick the 5 year one.  I HOPE that this means I have a higher quality product.  That is preferable.  However, that may not be the case.  It may mean it is the same product with extra warranty insurance built into the price.  I get that.  That is still acceptable to me.

Either I got a better product and I don't have to make a warranty claim within 5 years.  Or, I got a mediocre product and I do have to make a warranty claim within 5 years, but I don't have to pay for it.  What I don't want to do is pick the 2 year drive, have to replace it in, say, 3 years, AND have to pay for it again.

You might say, just buy 2 of the WD green and be done with it.  That is a potential option, but the math doesn't work.

A glance at NewEgg shows:

WD 1 TB Black, 5 year warranty, $ 88
WD 1 TB Green, 2 year warranty, $ 68.

So, two Green's would cost $ 136 for 4 cumulative years of promised service.

So, I'm still better off going with 1 Black for $ 88 for 5 years of promised service.  Even if I have to pay $ 20 for servicing two warranty claims on the Black, I'm still better off with the 5 year drive.

-snip-

>
>
>
>bnm



--

Sent from my Android Acer A500 tablet with bluetooth keyboard and K-9 Mail.
Please excuse my potential brevity if I'm typing on the touch screen.

(PS - If you email me and don't get a quick response, you might want to
call on the phone.  I get about 300 emails per day from alternate energy
mailing lists and such.  I don't always see new email messages very quickly.)

Ron Frazier
770-205-9422 (O)   Leave a message.
linuxdude AT techstarship.com
Litecoin: LZzAJu9rZEWzALxDhAHnWLRvybVAVgwTh3
Bitcoin: 15s3aLVsxm8EuQvT8gUDw3RWqvuY9hPGUU




More information about the Ale mailing list