Websites that Track PC Hardware Failure Rates?

scarolinus asks: "Working in the computer field, I find myself asking, "Which component to specify for my customers?" It used to be I would specify the most expensive component the customer could afford. The more established and reputable the brand, the better. Now, with no change in volume, it seems that failure rates are at least double what they used to be in the last few years. And it doesn't matter if the system is a $999 special or a $9999 multiprocessor server. So, is there a website out there that cataloges the failure rates of individual PC hardware components? Is there a website that helps us hold the hardware manufacturers accountable for the quality of product they produce?"

Comments

Service

[NetJunkie]

It would be hard to find out a components failure rate. The only person that has that information is the company that does the fixes and/or returns.

I think a better measure would be service and how the returns are handled. You could ocmpile that data from customers. To me, it isn't a big deal if a piece of equipment dies as long as it is replaced/fixed quickly. I know for my servers I keep spares around just in case... If I can't keep a spare for it I'm sure to have a 4 hour response time service agreement.

Re:Service

[Anonymous Coward]

To me, it isn't a big deal if a piece of equipment dies as long as it is replaced/fixed quickly.

1) That really depends on wheter the failure rate is once every 3 years or once every week.
2) Some people cannot afford failures at all.
3) There is always the possibility of (undetected) data corruption due to failure.

If I can't keep a spare for it I'm sure to have a 4 hour response time service agreement.

But what do those 4 hours downtime cost you? In production lines 4 hours can cost millions!

Storagereview

[cperciva]

Storagereview does this for hard drives.

Re:Storagereview

[slaker]

Note to folks who might participate in SR's reliability survey: If you do, please do everyone a favor and list all the drives you possibly can. It's a bit time consuming but the results thereby become that much better.

The SR community will thank you for your contribution.

Speak Danish?

[p0ppe]

The danish computer shop itbutikken actually list the percentage of products of different brands that has been returned to be fixed, when the percentage is high.

Large amounts of the page is, however, in Danish, but look for soemthing like this: "Info: Vores reparations afdeling kan konstatere at Net-hub af mærket D-LINK har haft en reparations frekvens på 0,82%. Tryk på det lille i ved produktet for at se om lige det produkt du ønsker måske har stor reparations frekvens. " ie. 0.86% of D-Link's net-hubs has been returned for fixing.

Re:Speak Danish?

[p0ppe]

Typo... ie. 0.82% not 0.86%.

Re:Speak Danish?

[agdvc]

en reparations frekvens på 0,82%[...]0.86% of D-Link's
Damn, Danish is complicated. Not only do you need to translate the words, but the numbers need to be translated too!

Re:Speak Danish?

[Anonymous Coward]

Gee, for HDs the number are >4.5% for the manufacturers listed !!

MTBF is what you're looking for

[hitzroth]

Many manufacturers publish mean time before failure (MTBF) specs on thier hardware. Try a google search on MTBF and whatever piece of hardware you're looking for if the data isn't available on the manufacturer's web site.

Re:MTBF is what you're looking for

[sigwinch]

Many manufacturers publish mean time before failure (MTBF) specs on thier hardware.

Most MTBFs are calculated values, derived from historical failure-over-time rates for the individual components inside the equipment. Take them with a grain of salt, especially if the equipment is used more heavily than was assumed for the MTBF calculations.

The other problem is that modern designs tend to push the components very hard, especially regarding heat dissipation and electrical currents. Even if the design is good, the components are still pushed into a regime where the historical data isn't as valid. And all too often the design turns out to be not good enough. (How many laptop batteries have been recalled over the last year?)

Usage of components.

[saintlupus]

Something to keep in mind for this sort of data compilation is the use that the components are being put to. For example, the dreaded IBM Deskstar 75GXP drives that are dying by the dozens this year. I've got one in my iMac, and it's working fine. I don't know a single Mac user who is having a problem with this drive. On the other hand, a ton of them are dying in AMD and Intel systems.

Is it a heat thing? An OS thing? Damned if I know. But if you polled these two communities, you'd probably get vastly different results.

--saint

Re:Usage of components.

[dublin]

Reportedly, the earlier ones worked well, and it's primarily the later ones that are a complete disaster.

Sad, because traditionally, IBM drives have been head and shoulders above their competition in reliability. (This is *especially* true of laptop drives.) I switched to Maxtors a few months ago, partly because of the IBM 75GB problems, partly because I wanted 100 GB drives anyway, and so far, the've performed beautifully in non-trivial quantities in arrays - not one failure so far. I'm looking at the new 7500 RPM WD drives, but am reluctant to consider giving up the reliability of the Maxtors...