Hardware failure rates


Father Xmas

 

Posted

The original quoted article is in French so I'm not sure how in-depth or rigorous the numbers are, but this article has some interesting anecdotal numbers:
http://www.macworld.com/article/1563...ssdsurvey.html

Some nuggets:

Quote:
If you're looking for an ultrareliable power supply purchase, the Thermaltake EVO Blue 550W had zero returns...

In the top spots for individual memory product returns are a staggering 15.08 percent failure rate for the Corsair...
Original article here (en francais):
http://www.hardware.fr/articles/810-...omposants.html


 

Posted

This study does not cover what I was hoping it would. I could argue that they are stretching to making a claim based on the evidence. DOA (dead on arrival) rates do not reflect reliability of electronics over a period (years) of use. DOA's simply reflect quality issues in manufacturing that will always exist, no matter what type of electronic component you are examining.

In their first paragraph, they claim that solid state drives are less reliable than mechanical. In the first 30-90 days, this may be true due to new manufacturing processes for SSDs requiring optimization and quality improvements. However, over years of use, mechanical hard drives will not last longer than the average solid state drive due to wear from friction.


 

Posted

Quote:
Originally Posted by Syntax42 View Post
This study does not cover what I was hoping it would. I could argue that they are stretching to making a claim based on the evidence. DOA (dead on arrival) rates do not reflect reliability of electronics over a period (years) of use. DOA's simply reflect quality issues in manufacturing that will always exist, no matter what type of electronic component you are examining.

In their first paragraph, they claim that solid state drives are less reliable than mechanical. In the first 30-90 days, this may be true due to new manufacturing processes for SSDs requiring optimization and quality improvements. However, over years of use, mechanical hard drives will not last longer than the average solid state drive due to wear from friction.
True, but I'm not sure any organization could do a study like that on a large scale except for the manufacturers themselves, and most likely they will not release any data that makes them look bad.


 

Posted

Quote:
Originally Posted by PumBumbler View Post
True, but I'm not sure any organization could do a study like that on a large scale except for the manufacturers themselves,
Google did. They have a bajillion hard drives in their data centers. Some of their engineers published a paper, "Failure Trends in a Large Disk Drive Population". PDF of the paper is here. It only covers hard drives, but it's something.


Paragon City Search And Rescue
The Mentor Project

 

Posted

Quote:
Originally Posted by Ironblade View Post
Google did. They have a bajillion hard drives in their data centers. Some of their engineers published a paper, "Failure Trends in a Large Disk Drive Population". PDF of the paper is here. It only covers hard drives, but it's something.
They don't mention any specific model or vendor in that study, which makes choosing a specific recommendation difficult.


 

Posted

Their point was it didn't matter who made the drives, failure rates directly correlated with time spun up. All other factors were statistically insignificant compared with that.

Still if you looked at the article from hardware.fr, it's not failure rates really but returns based on one french online retailer. Take memory for instance. Corsair overall only had a 1.4% return rate while a particular model had 15%.

In the case of memory how many of these returns really come from dissatisfaction when the buyer attempted to use them in an overclocking situation and was unhappy that it didn't perform the same as some review site.

Tom's Hardware just got bit by this attitude on their new $2000 build. They went with very cheap memory that in the past they found they were able to OC well above it's listed speed and timings but now, OMG they actually meant DDR3-1333 Cas 9 when they say DDR3-1333 Cas 9. What is the world coming to? That's OK, they saved a whole $15 over DDR-1600 Cas 8. But that $250 case really helped performance.

Same is true with hard drives, video cards, motherboards, etc. A review shows a lucky result overclocking or fortuitous combination of parts and when they try it themselves their results bares little resemblance to the results from the review they read. So they send it back with a sob story.

Return rates to an e-tailor isn't actually failure rate but skewed with a dissatisfaction rate due to over hype.


Father Xmas - Level 50 Ice/Ice Tanker - Victory
$725 and $1350 parts lists --- My guide to computer components

Tempus unum hominem manet

 

Posted

Quote:
3.2 Manufacturers, Models, and Vintages
Failure rates are known to be highly correlated with drive models, manufacturers and vintages [18]. Our results do not contradict this fact. For example, Figure 2 changes significantly when we normalize failure rates per each drive model. Most age-related results are impacted by drive vintages. However, in this paper, we do not show a breakdown of drives per manufacturer, model, or vintage due to the proprietary nature of these data.
I am reading the successive part as being that normalizing within each brand/type as showing the spin up time as being a valid metric for reliability, but it makes no claim about whether that is overwhelmingly the metric to judge reliability versus any failure rates from a specific model or vendor.

Also, your claim about returns being made because of hype or disappointment can be true but does not rule out absolute failure either. That is only a supposition, but that in itself is not unreasonable.