Slashdot Mirror
Failure Rate of PC Manufacturers?
The ever-popular Anonymous Coward asks: "Hello.
We are conducting a write-up for our clients, however we cannot seem to locate any published failure rate of PC manufacturers. Google does bring up past PC Magazine articles - but nothing recent. Does Slashdot know of a way to find this information, as this strikes me as valuable information for the computer buyer. We sell many PC's (B3 VAR) and have done for the last 5 years. We can and will produce our failure rate info - why aren't the big companies doing so?"
...but at these prices, who cares about failure rates?
:) about people buying whole new systems just because their current PC is loaded with spyware.
For personal use the PC will most likely become obsolete (at least in the eyes of the user) before it becomes broken. On slashdot we've seen stories (over and over again
If it's for business use, and you've got to have 100% uptime, failure rate sill doesn't matter, since at these prices you buy multiple redundant systems and then sleep well at night.
Besides, how do you collect your data? It seems to me that by the time you've got good long-haul use data on your systems you won't be selling them anymore in favor of new models. And I don't see how extropolating data for new models based on old model performance is terribly useful.
By way of analogy - if new cars only cost a grand, you'd replace your car long before anything serious went wrong with it. About the time the ashtrays were full, a flat tire would be just the excuse you'd need to go shopping for the Latest Greatest Leetest Carxen.
It's a "gentleman's agreement" among the top-tier PC makers. I won't make you show yours if you don't make me show mine. There is very little to gain, and a lot to lose, especially if you consider how malleable the definition of "failure" is in the PC market. User deletes system files and renders PC unbootable? Chalk it up as another failure!
Blogging Weight Loss, Distance Education, and more at verlin.com
Consumer Reports publishes rankings of reported defects by brand for a zillion product categories, including PCs and, I think, printers, scanners, whatnot. It comes out once a year or so. I have a subscription, so I don't know how non-members can get access, but the magazine and web access are relatively cheap. They only cover the top brands, and they only report what their surveys have collected, so it may not be as double-blind and objective as one could wish. It is probably also not directly comparable to your data.
Infuriate left and right
the failure rate of a machine is more dependent on the environment. if it's extremely dusty, fans clog and processors overheat. if you restart the machine a lot, it will probably fail sooner than if you leave it on all the time (or so they say). if you expose it to children (or noobs) it will fail sooner.
how do you compare an [insert high quality mfgr] box located in a kintergarten library (dusty + children) to a no name box used by post docs in a climate controlled research lab?
all the major manufacturers will use similar components. it's like lego!
I wonder, if this information were to become public, how long after it's out of the box are we talking about? First boot? 1 year? And how long until it's obsolete?
Personally, I've found that Compaq/HP systems more often than not have problems right out of the box (or right after the system restore, as it were), while Dells have a pretty good run, until the end user mucks it up with malware. I haven't had too much experience with Gateway systems to get an idea about their outcome.
Those were the days. PC Magazine would review 50 PC's, from ATR to Dell, Polywell to Gateway. Big names, small names. They ran application benchmarks and you could see which systems were the fastest. They also included information such as failure rate and wait times for calling tech support.
These days, with 'web magazines', a PC comparison has 5 PC's and a paragraph or two about each one.
Of course, magazines weren't perfect. The top rated PC's were often the most advertised. Manufacturers probably got smart and decided to pressure magazines into not publishing those numbers by witholding advertising dollars.
Based on more than 69,000 desktop (73,000 laptop) computers purchased new from 2000 to 2004. Data were standardized to eliminate differences linked to age and use. Differences of less than 4 points are not meaningful.
Repairs and serious problems:
Desktops:
Laptops:
We need a new four letter acronym:
Read
The
Fucking
Summary
And not only the title.
They have reports that cover this sort of thing. It costs money, but they get the data straight from the horse's mouth, as it were.
Of course, if you're just a podunk little outfit that they think will redistribute this stuff at a drop of a hat, they may refuse to sell it to you. But it can't hurt to ask.
-EvilMagnus
Come on, Gateway, almost 1 in 4 desktops with repairs or serious problems? Compaq/HP, IBM and eMachines with around 1 in 5? The numbers don't look any better on the laptop side, once you factor in that a broken laptop generally has fewer parts a user can easily replace than a desktop.
If I ever needed ammo to prove that building it yourself is better whenever possible, this is it. Although I'd like to see motherboard failure rates, also, and not just reported by the manufacturer, because hardly anyone really returns them after the store refund period passes.
The generic answer seems to be, from my experience, that the chance that some computer will fail to the point of not POSTing during their warranty period is upwards of 60-70% with corporate desktop machines (think Dell Optiplexes and the like). The Gartner reports could be helpful if you can afford them, but what might also be useful is if the articles you find from the past show a downward trend in reliability from the big boys (which seems be the case from my experience, at least), and a steadily higher or improving trend in reliability from your comp.
PCWorld Magazine (and their corresponding website) hold an annual survey of reliability and customer satisfaction of the major PC brands.
No link here, as it's been a few years since I have subscribed to the (dead tree version of the) magazine. Google is your friend, though.
I'll create an amusing sig when I have something meaningful to post.
emphasis on average
sometimes we had items with 10% failure, but usually most things were like 1 or 2%
not that it seemed like that when working helldesk
---- Put Sig here:
Check any issue of Consumer Reports that reviews PCs. They have a chart ranking the relative failure rates of the different manufacturers.
Nobody wants to publish failure rates - even if they're low, it still sounds bad. Try looking for "success" or "satisfaction" rates instead! Remember, marketing people just don't think like we do...
(back when I worked in the repair depot, I remember Packard Bells were approaching a 50% failure rate. Then they merged with NEC - still not sure who got the worst of that deal, given the crap PCs NEC used to ship...)
Perfectly Normal Industries
For the general desk top including small office servers, find a local white box builder who can churn out systems. Specify motherboard and CPU. No the shop won't have 7x24 coverage, but then neither does Dell, HP or Gateway. If you can drive down the block and get a system serviced or swapped out you'll beat the hold time before your problem is addressed with any of the major vendors.
For a server class you need build enough redundancy into your equipment to have a box down and still provide services. Until you get into the major application white box is still the way to go.
Actually I did read the entire article summary and still thought the entire time that it was talking about PC manufacturers going out of business. I guess I am too dense and I have never thought of "failure rates" of PCs in that sense. Oh well.
Except that the HP failure rate is too low (i.e I was seeing 50% failure out of the box.)
Toshiba is too high. I've been lucky and seen 0% there.
I wonder why NCR didn't make the list? We were seeing over 80% failure there.
When I worked at Tandem in Austin, all systems were built, run through a complete set of diagnostics and soak testing before shipment. The systems were designed as fault-tolerant and sold mostly to the telecom industry. The goal was zero defects and to NEVER ship a DOA system. Tandem's systems were expensive, and the company competed on quality, not on price. Need I add that Tandem no longer exists as an independent company. (Bought by Compaq, which was bought by HP).
[Insert pithy quote here]
If they don't have a published failure rate, you can get a good idea of the system's failure rate by looking at the MTBF (Mean Time Before Failure) numbers for the individual componets (except for any proprietary hardware).
If it's for business use, and you've got to have 100% uptime, failure rate sill doesn't matter, since at these prices you buy multiple redundant systems and then sleep well at night
If you need redundancy, it's probably at a server level. Servers are not necessarily available at such cheap prices.
For desktops, if the unit dies in 3 months then you can probably get a better unit at less cost, but you still have to have somebody paid to swap the machine/components and reinstall software as necessarily.
If cars cost a grand and were shyte, I'd buy a car that wasn't shyte and cost more...
Back in 1997 when I worked at Computer City, a Compaq rep informed us that they were running at around %35 DOA (dead-on-arrival). Nice to see they've gotten that down to %21. I'm surprised Dell isn't worse, considering all of the GX270s failing left and right at work.
When I ran a PC manufacturing company some years ago we would release those figures if anyone asked for them. Pretty pointless, as people had nothing to compare them with, and as has already been pointed out, the figures are influenced by the user and the environment.
What do you consider a failure?