Ever deepening hatred for WD

[Radboy]

http://www.drugstore.com/qxp40608_333181_sespider/leight_sleepers/ear_plugs.htm

 

[jtr1962]

I don't care that the reliability results for all drives are off by a factor of ten as long as all drives are more or less off by a factor of ten. It's the relative differences I care about. And I think that even the SR reliabilty survey preserves enough of these differences to be able to say that the X15 (98th percentile) is more reliable than the Maxtor DM+9 (21st percentile).

This assumes that whatever reporting biases are present in the survey are spread across all makes of drives. It also assumes that the samples for each drive are large enough to draw reliable conclusions. While you might make a good case for the first part, the number of people self-reporting for each different type of drive simply isn't enough in many cases to give decent results. If you remember from your statistics courses the margin of error gets smaller as the sample size increases. If you flip a coin 1000 times you will almost always turn up heads between 49% and 51% of the time. Not so if you only flip a coin ten times. You may turn up heads 3 times or 7, and the chances of that happening are fairly high. Same with the SR Drive Reliability survey. If only 10 of each size and type of drive are in the database, I personally don't think I can draw any conclusions worth a damn from this. It may be better than nothing, but not by much. The only meaningful conclusions I think will come from the manufacturers of the drives themselves, but that is most likely privileged information that will never see the light of day. Data from someone selling large numbers of drives of the same type is almost as good. Presumably any failures due to mishandling will be spread equally among all models since it is the same customer base, so if the sample sizes are large enough any differences in RMA rate will be due to differences in reliability.

On a personal note, I have yet to have a hard drive fail in any of the machines I own. You might conclude that all makes and models of drives I own are therefore 100% reliable, but due to the small sample size the results would be meaningless. Actually, every drive I've purchased has been a Maxtor, but I've inherited a few second hand WD and Quantum models as well. Subjectively, WD always is noisier (i.e. more whine) to my ears. Perhaps the dB level is the same as other makes, but there seem to be more high frequency components to which my ears are very sensitive.

 

[CityK]

http://www.drugstore.com/qxp40608_333181_sespider/leight_sleepers/ear_plugs.htm

Earplugs always become uncomfortable for me after short durations.

I'd love to, for example, get ahold of Newegg's RMA rates for hard disks, instead.... now that I think of it, has anyone ever asked a company like newegg for that information?

That's a pretty good idea. Newegg would be great from a large sample size perspective.

One drawback might be the question of how many of those defective drives would be resultant of shitty Newegg packaging jobs (as many people have reported Newegg could use some lessons on how to properly package a drive up for shipment).

Also, would a large retailer have to enter into any sort of NDA about releasing any kind of failure information about the products they purchase?

 

[ddrueding]

Also, would a large retailer have to enter into any sort of NDA about releasing any kind of failure information about the products they purchase?

You bet. You'll never see this info.

 

[Bozo]

I questioned the vender we used for computer parts about hard drives. They would not give me any numbers, but I was told that they have far fewer WD drives returned than other brands.
I have bought Seagate, WD, IBM, and Maxtor from this vender.

Bozo :mrgrn:

 

[ddrueding]

The largest truely controlled statistical source I know of is Tony...but it isn't exactly comprehensive.


Hmmm....Samsung good.....

 

[Adcadet]

for reliability info you can just go to SR and use their Reliability Database. What more perfect data source and analysis engine could you possibly want? Don't tell me you don't believe their results.

 

[e_dawg]

This assumes that whatever reporting biases are present in the survey are spread across all makes of drives. It also assumes that the samples for each drive are large enough to draw reliable conclusions. While you might make a good case for the first part, the number of people self-reporting for each different type of drive simply isn't enough in many cases to give decent results. [...] If only 10 of each size and type of drive are in the database, I personally don't think I can draw any conclusions worth a damn from this. It may be better than nothing, but not by much.

Yep. I did mention that newer drives with few samples reporting were not as trustworthy as older drives with a large number of samples reporting.

The only meaningful conclusions I think will come from the manufacturers of the drives themselves, but that is most likely privileged information that will never see the light of day.

Yep.

Data from someone selling large numbers of drives of the same type is almost as good. Presumably any failures due to mishandling will be spread equally among all models since it is the same customer base, so if the sample sizes are large enough any differences in RMA rate will be due to differences in reliability.

I disagree. Sometimes it's true, but the problem is that different drives sold by the same retailer can take different paths through the marketing channel through different distributors, different couriers, and different handlers. And that's not all. What about differences in packaging? For example, Seagate's SeaShell clamshell packaging (often used even for OEM drives) offers their drives greater protection during their journey through the channel than the plain ESD baggies used for some of the other drives. Seagate has data that shows their SeaShell packaging makes a significant difference in the number of returned drives. Packaging standards are not uniform either. A nationwide survey like SR's is more likely to randomize any such effects from packaging, shipping and handling differences.

 

[Adcadet]

I disagree. Sometimes it's true, but the problem is that different drives sold by the same retailer can take different paths through the marketing channel through different distributors, different couriers, and different handlers. And that's not all. What about differences in packaging? For example, Seagate's SeaShell clamshell packaging (often used even for OEM drives) offers their drives greater protection during their journey through the channel than the plain ESD baggies used for some of the other drives. Seagate has data that shows their SeaShell packaging makes a significant difference in the number of returned drives. Packaging standards are not uniform either. A nationwide survey like SR's is more likely to randomize any such effects from packaging, shipping and handling differences.

To help convince people that confounding isn't a big issue people (journal authors) usually give some info about potential confounders (not that there is a test for confounding, anyway). Last I checked SR has never done anything like that. Have they ever given any demographic info about the data in their database? For all I know every WD drive in their database was sold through Vendor X who puts every bare drive through a clothes dryer before shipping.

 

[e_dawg]

For all I know every WD drive in their database was sold through Vendor X who puts every bare drive through a clothes dryer before shipping.

Cheaters. The centrifrugal forces imparted by the dryer help align the magnetic particles on the drive and ensure proper electron spin. WD drives have an unfair advantage!

 

[sechs]

It's like stonewashing. It adds style by wearing them out before you get them.

 

[flagreen]

Tannin, would you mind not answering my questions?

Ahem ....

Bozo: 4 failures out of 75 drives is not an acceptable failure rate these days. That is about 5.3% You should be getting a failure rate about ten times lower than that. Our Samsung failure rate currently stands at less than 0.3% and has been stable on or around that figure for quite some time, on a sample that is now gradually approaching 3000 drives. (7 failures within the three year warranty period, I haven't counted the total number of drives shipped lately, but over 2500, I think. I'll try to find time to check exactly sometime soon.)

A sample of 75 isn't definitive, but 4 failures is sufficient to give a reasonable indication, and your Western Digital drives are failing more than fifteen times as often as our Samsungs.

That5.3% number doesn't really surprise me, as our own WD failure rates (back before we stopped selling Seagate and Western Digital) were somewhere in-between your numbers and Mercutio's.

Good Lord! Don't tell me that Tannin is relying on his own failure rates to make decisions as to what Hard Drives he sells! How unscientific can you get? He might as well be relying on SR's Reliability survey.

 

[time]

At the risk of sounding anally retentive Bill, it's not a self-selecting sample.

ALL of the drives he sold are included, rather than a sample selected by the whim of individual customers.

The only real issue is the fact that he buys them from a single supplier.

 

[flagreen]

At the risk of sounding anally retentive Bill, it's not a self-selecting sample.

ALL of the drives he sold are included, rather than a sample selected by the whim of individual customers.

The only real issue is the fact that he buys them from a single supplier.

As you may have guessed Dave I know little to nothing about statistics. Thank you for the explanation. I'm glad it was only Tea that I was teasing and not Tony. :mrgrn: (Mums the word Tea).

 

[Bozo]

I've been checking some of our hard drive failures. This is not 'statisical' or 'scientistic' but does offer a closer look at what failed.
First off: this a heavy manufacturing plant. Most of our eqiupment converts AC current to DC current. Some convert DC back to AC and pump it back into the AC lines. Besides having a low Power Factor, our electric is very dirty. The computers that I'm working with are for process control. Some are on a UPS, others aren't. All these computers run 24/7.

Maxtors: We had 13+ Maxtor hard drives. All of them ran for 5+ years without a failure. Most have been replaced now, but all were running when changed out. Some of these have since been installed in non production computers.

IBM: We had 8 Deathstars of various flavors. All ran for almost three years. None failed. They were changed out because of their reputation for failing. These are still used in test setups.

Seagate: We have 28 Cheatahs, 15K rpm. All have been running for 2 years. Two have failed.

Western Digital: 68+ installed. This is a mix of BB and JB drives. Most have been running for almost 3 years. We have had 3 failures. All of them the BB series drives. All were from computers in non-climate controlled, non-UPS areas. We also have 18 SATA drives, including 8 Raptors. No failures, but these have only been runnig ~8 months.

Bozo :mrgrn:

 

[Computer Generated Baby]

Western Digital: 68+ installed. This is a mix of BB and JB drives. Most have been running for almost 3 years. We have had 3 failures.

At work (of course), circa 1998, a group of office worker types had something like 38-each Western Digi 6 GB hard drives with approximately 30 failures in fairly rapid succession, with the remaining WD drives still working at the time of removal. All 38 or so WD drives were replaced by a couple of different cheap Maxtor drives. I believe all of those Maxtors lasted until those computers were replaced 2 or 3 years later.

At work, again: Some dedicated workstations for image scanning, which required 2 separate physical ATA drives -- no SCSI hard drives allowed, were spec'd with what was the latest and greatest ATA drives. The ATA drives in question: IBM Deskstar 75GXP 30 GB and 60 GB models. Believe it or not, all er... 12 of those hard drives are still working perfectly! Go figure.

 

[Bozo]

About the same here: In 1998, the company bought over 300 Gateway E4200 computers for the office types. All had a WD Caviar 6GB hard drive in them. No one in IT can give me the number of hard drives that failed, but they don't believe it was all that many. Some of these computers are still running today.

Bozo :mrgrn: