Slashdot Mirror
Windows Fails 8% of the Time
descubes writes "A Journal du Net article reports that about 8% of Windows sessions require a machine reboot.
The relevant quote (translated from french) is: "The average rate of failures requiring a system reboot has been measured at around 8% per session. This number varies widely depending on the version of Windows. Windows 2000 has a failure rate of 4%, and NT4 is at 3%, whereas Windows XP is close to 12%." The study was originally made by Acadys and Microcost and gathered data from 1.2M machines belonging to about one thousand companies over a period of one month in seven different countries."
And what is the reboot rate of various Linux distros? Unless they're willing to do a comparison under the same protocols, I very much hope that no one here points to this as more proof of needing to switch to Linux, even though I know it will come up.
The Braying and Neighing of Barnyard Animals Follows.
One does need to wonder under what conditions those computers were in. My Windows XP boxes hardly ever crash, and if so usually its a hardware failure (Video card overheats, processor overheats [welcome to Florida!]). All the computers we have at the college run Windows XP, specially tweaked to keep students doing school work [not dorm boxes] and will clean themselves up when they are rebooted.. these boxes too usually never fail unless its hardware, and operate all day with many different users per day. I also wonder, since my views are somewhat cleaned by our nice IT folks at the college, what these computers they monitored were like. Was there ad-ware on a few? A few viruses maybe? It happens, and IT can't always be there to fix those problems.
My point simply is usually its not Windows XP faulting for me, its something else not getting along with it. Be it [insert]ware, or hardware issues. Good example is I hardly ever reboot this computer, it has easily gained weeks of uptime, usually only shutting down due to thunderstorms taking out the electrical lines.
I've left to find myself. If you happen to see me, please, keep me there until I return.
... is that the windows faliure rate is INCREASING.
This is about rebooting. A crash is not the only time Windows forces you to reboot. You say you shut down daily - only Windows users would regard that as normal.
remember XP doesn't "crash" anymore, it just reboots, which could be excused as just some random power glitch and may go by without thinking it's actually windows fault. after all, some times linux just reboots when the lights dim too. XP doesn't need the lights dimming, but then, maybe i just didn't notice it "this time".
if you turn the BSoD back to "stay on error message" rather then "just reboot" you might start to notice XP crashing again.
What's a 'session'? They give me XP at work. Not my idea of a good time. I reboot XP when I don't understand what's going on, but usually I don't know if XP has failed. It seems to have some problem with degradation of the management of some resource (maybe memory) over very long sessions (a week or more). Then, when the machine gets sluggish and recalcitrant, I reboot. But maybe it's just the network admin spying on my machine or something that I don't even see. Damfino.
what's so wrong with shutting down daily?
If you have your PC at home, you shut down at night so you can save electricity, and stop the noise from the fans.
If you're at work, sure you have less incentive to shut down, except to save electricity again. (save the planet, man)
Personally, I leave my w2k work box running all the time, but even then it gets shut down for the weekend.
"Windows works great, for people who know how to use them. (Same can be said for Linux, Mac, etc)."
You've hit the nail right on the head, and done it without any OS-based zealotry.
Let me be the first to say "180 days? Wow!!!".
Just kidding. Although I do love the story about the Novell server at some University (in Florida, I believe) which had been running for several years with no reboots and no problems. One day some brilliant tech decided to look for the server and realized that it wasn't there. Nowhere to be found.
Fast forward a couple more years, they were doing construction, and found the server had somehow been put in a closet that had been bricked over - meaning that the server had been running without intervention for close to 5 years without a reboot or software updated. Go Novell! Running on Compaq hardware, IIRC.
seriouslyexcited.net
in my CS department. The amount of crow that is getting passed around is amazing these days as many are being forced to switch to Linux or MacOS X for class in the 400 levels and they realize "uhhh those UNIX guys were right about Windows." The irony of it is that we Mac users are usually very good at helping them get started with OSX.
Still, we can't blame Microsoft for a lot of the instability since there are so many users out there using terrible and/or outdated drivers. Microsoft cannot be blamed for the quality of the drivers that most Windows users have because they didn't write them.
Of course I will say this about Windows. It is nice for the first few months, but then it just begins to become as sensually appealing as a rotten piece of bait fish left on your back porch for a few days in the sun. My Macs frequently have several times the uptimes of the Windows PCs I hear about and the Windows users are shocked, "why are 8 weeks of uptime, your PowerBook is still fast and usable."
Click here or a puppy gets stomped!
I have (and have had in the past) many Win2K boxes and Windows XP boxes that run 24/7 except for forced reboots from software patches and/or hardware installation.
One Win2K server I helped maintain had a better uptime than most of our Suns and other Unix boxes. Mostly because it was well protected from the 'net and we didn't patch it.
As far as other OSs, my Linux servers also run 24/7 and have a high uptime. However, as of late, we've been notified (by the various distribution patch notification tools) of more software patches than on Windows by a long shot. My SuSE 9.1 Professional box, for instance, hasn't gone a week without at least two patch sessions for the last two months, but only a few of those have required reboots (kernel patch).
Well I set up a Intel Raid drive setup under XP Pro and have suffered several blue screens of death. Another issue I am having is XPs DVD-RAM driver won't recognize some FAT32 disks that were formated under 2000, this happens at random some disk work others don't, never had this problem under 2000. So far from perfect...
Onward to the Aether Sphere!
THis is a bit misleading, I think. I run some of the most crashworthy programs you can imagine on XP. 3D apps with beta drivers, AVID editing software, After Effects-- all things that are known for their crashiness, but it's VERY rare that I have to reboot. I do, howevever, ocassionally have to kill a process. Many users may not know how to find the misbehaving process and kill it. So they do what they know how to do-- hit the reset switch.
Ocassionally, while running Doom3, I might hard lock-- My office isn't well insulated, and my machine can get pretty hot when stressed.. Plus I'm running hacked drivers on my video card, so I don't really blame anybody but myself. Otherwise, I cannot remember the last time I HAD to reboot other than software/driver installation.. (And driver installation doesn't always require that anymore...)
This level of stability, in my experience, is virtually the same in Linux.. It runs programs that ocassionally crash, or you have to kill em, and you can get hardware video lockups causing a reboot if you try to do "daring" things (which most people don't do because of the lack of games/3D apps for linux.) I'm not trolling here, just trying to objectively compare the situation..
I think this is just part of computing-- and maybe all OS'es can do a better job of recognizing what apps are really crashed, and helping the user dispose of them a bit better.
Well, seeing how they calculated crashes per session, your crashed session rate would be about 20% (1 crash per month, 50 work weeks in a year). It is odd to calculate it that way, since with more reliable machines you might leave them on until they crash or you power down, leading to higher numbers. The most reliable system could have a 100% rate of crashed sessions.
It might be more appropriate to keep track of how often people need to reboot.
Information: "I want to be anthropomorphized"
Well, Beavis, I can't argue that XP looks retarded, but I don't know that there's much difference to customizing XP and say, getting KDE just right, or worse, having to download, compile, and configure XFCE (which is awesome).
I'm basing my assertion on experience. I am the IT manager/network admin of hundreds of computers, and I used to work for the Navy, where we had a very large IT department. In my experience, XP is extremely stable. Lots of times you can do simple things like restart a service to "repair" the system without a restart. There's no doubt in my mind that thing that Linux users do would NEVER be done by a Windows user.
After all, if my Linux box has a problem, I'll drop out to the terminal (Al+F1), login, su to root, kill -9 the hanging process, shutdown offending services/threads, and manually relaunch them.
On XP, Explorer and other process will restart themselves if they crash, but sometimes you have to wait. And other times it needs a push. I'm willing to bet the average XP user, at first glance, reboots. Does that count?
That's right.
My own experience is almost never having to reboot on WinXP Pro. I have 3 machines running all the time and except for some hardware or software installs that may require a shut down or warm reboot they just run 24/7. My media server is XP pro and it never dies - and I'm running it on an old POS P3 with 512 RAM. Myself, I think XP Pro is extremely stable - not perfect, but stable.
I think casual users versus knowledgeable users is a big factor here in the study. A properly managed box can run for a really long time without interference. Some may claim that the degree of knowledge required to "properly manage" a WinXP Pro box is too high a threshold for the casual user - so that might be true, but the same is true of Linux and most OSS.
Personally, I don't see that as a totally bad thing. People should have a solid understanding of what is going on with their boxes. Once upon a time I knew far less and my satisfaction as a computer user suffered for it - I was a clueless idiot and bad stuff was always happening to me because of it. Now I know at least something about what I am doing and I am a far happier user - stuff almost never goes wrong.
Somebody should do a new study on knowledge as relates to satisfaction of use of a product. I just bet the more you know about something the happier you are in relation to it. Satisfaction comes from knowing the product inside and out - because while you may become familiar with the product's shortcomings you will also come to know how to easily work around those shortcomings to get what you need out of a given product.
I wonder about format intervals. I know that after 6-18 months, my XP box can degrade to the point of requiring manual reboots constantly. A reformat/reinstall typically brings me back to ~95%.
G
Having played with win98 for the first time in nearly five years, and getting a BSOD during the install process, yesterday, I can say that you are correct. XP/2k actually work as the MS marketing dept says they will... mostly... mostly... they come at night... mostly...
Does it go on forever?
The XP box on my desk at work has never crashed by any fault other than my own (testing funky code), but the higher end "gaming" box at home has crashed a number of times. Usually while playing a game.
Personally I blame the craptastic drivers from both nVidia and ATi. They're hell-bent on getting the most flips per flooble and the stability of the drivers suck.
ATi adds a lame new interface (which crashes) called "Catalyst Control Center" while the actual usability of the drivers is swirling down the toilet. All new releases focus on little tweaks in their $500 dollar range cards to make it benchmark fastest in Doom 3, while support for the cards people actually own dwindles.
For instance, if I try to play Doom 3 with anything higher than "Medium" settings, my machine will hardlock. Radeon 9800, no tweaking or overclocking, just the latest "stable" drivers.
This isn't an anti-ATi rant, I had the same bullshit with nVidia.
Barring a hardware faulure, it's virtually always the video drivers fault, since it actually has the power to bring down the system.
I'd say the higher instance of XP bombs reflects it's status as the current PC gaming platform.
I blame nVidia, ATi, and Microsoft for "certifying" their instable, shit drivers. Driver certification really just means your check cleared.
What can they do about it, though? I'd gladly sacrifice a few FPS for a stable machine. But when a driver release gets less "3DMarks" than the one before it, the little kids throw a fit on rage3d and other sites.
I don't need no instructions to know how to rock!!!!
Even more so with the new fluid dynamic bearings in newer hard drives. The only time they actually wear is when they are started and stopped, otherwise there is no actual contact between the bearing surfaces and so they can theroetically last forever. I read about a 100 year old hydroelectric turbine that used the water as a fluid bearing that had no significant wear and was predicted to last at least another 1000 years.
Smoking is an expensive, slow, and unreliable method of suicide.
Infected in 20 minutes
Out of the box home windows xp has on average 20 minutes (if on a uni network, much less) before it is taken over.
corporate networks should all now be firewalled... shouldn't they?
#hostfile 0.0.0.0 primidi.com 0.0.0.0 www.primidi.com 0.0.0.0 radio.weblogs.com
I'd be interested to know just what you were doing when it crashed. My system tends to be stable unless I'm farting around with low-level stuff or drivers..
My XP box is pretty stable, but I have had it blue-screen (although it is a nice shade of blue in XP), a few times, doing fairly normal things..
I update regularly, and would say that XP seems more stable that any version of windows I've used. But I don't really use it that often, only the odd time when I must use windows and wine does not suffice.
Even in corporations where there is an IT staff, I see computers brought in all the time full of spy-ware and viruses so, you can't say that since there is an IT staff there won't be any problems. In our own environment we used 2k for a few years, and now use XP. I never reboot my computer unless absolutely needed, it runs 24/7. I do development with VS, flash, Photoshop, and several other environments. With 2k I would have to reboot at least once a week to keep things stable sometimes more than once in a day. With XP I only reboot once every 2 weeks to once a month. In my personal experience XP is a much more stable environment.
Well, download F@H and give it a purpose!
Yeah, I understand the point you're getting at. I personally leave my zd7000 running 24/7 because most of the time I'm leaving it to do a render overnight or something. Most PCs in offices ought to be based on Transmeta or Via low-power CPUs anyway, seeing all they do is web surf, word process, and remote sessions.
I wish I could write clever and witty sigs.
Crashing and requiring a reboot are two different things. I use XP at work too. I have ZERO spyware on it. It is for work, I use it for work only. No button bars, no cute apps. The only thing I use personally on it are Opera, PuTTY, and an old version of Winamp. I have to reboot about twice a week.
If people need to reboot 12% of the time, then they are doing something wrong. It's not the OS, but more the user in my opinion. XP is a stable system, and does a good job of keeping my machines running.
I have a good idea why my system needs to be rebooted, it is some of the apps I run - mainly certain Rational tools. Sure, on Win98 it would blue screen and crash. XP will just slow to a near halt or start behaving very oddly. Reboots are part of Microsoft OS maintenance. If there is a problem with your machine - reboot. SOP, everywhere I have been.
Even if XP is stable, if it allows applications to bring it down and make it unusable, then the PC isn't stable - period. If the OS can't control it, then it is the fault of the OS.
Hey, I have problems at home on my Linux machine too. Apps will cause X to freak out, and I have had to reboot because I don't know how to cleanly shutdown X remotely or from a console. I am sure there is a way, it just happens so infrequently I haven't bothered to find out. Sometimes Opera will crash X, or if I am messing around with settings on Mplayer, it will freeze it. I used to have problems with my Xfs (font server) crashing all the time, but that was on my old system (Redhat 7.3). I think that may have caused some of the problems with Opera freaking out. I just upgraded to Mandrake 10.0 a few weeks ago, so hopefully that is all straightened out. But my uptime at home is usually VERY long. Not to start comparing, but it usually gets rebooted only when the power goes out or something. In fact, my web server has been up since the last power hit, 118 days ago. Before that, it was up over 230. :-)
My beliefs do not require that you agree with them.
But in an enviroment filled with Google Bar, Webshots, Gator, Weatherbug and other crap, not including the pure spywear and viruses, the PCs will fail. It has nothing to do with the OS, but everything to do with stupid users, and a lazy and ineffective IT department.
Hoist Number One and Number Six.
I always log out of KDE at the end of the session (*), but my machine (home-office workstation) normally stays up until it has to be hardware-serviced or I want to upgrade the kernel. I maxed this in about 100 days. X never locked me up badly, at least not since Xfree 3.1 or (ie, a long time ago).
I'll have to agree there actually. Most recently even when a program has managed to lock X up, it still respects Ctrl-Alt-F1, from which I can kill the offedning program(s) and X bounces back happily. I guess this is the equivalent of Ctrl-Alt-Delete and using the Task Manager in Windows. The Linux method (while less user friendly) has the advantage that you drop right out of X, and hence have full control of your machine again. Trying to haul up the task manager when the GUI is locking can be rather difficult sometimes.
Jedidiah.
Craft Beer Programming T-shirts
Of the reboots that were done how many could be avoided by knowing how to get out of what caused the lockup? I know that the average user just does a reboot to get the problem solved when ending a task might get them out of a jam.
XP is a stable as hell, with the following qualifications:-
Not if you live by a aluminium smelter (power issues)
Outlook, worms, virus's and malware/spyware have decreased reliability. Also employee 'monitoring' software tends to explode or corrupt profiles. In one case, the help desk, and the image testing team, were not told security were sneaking in a extra payload - yep, the sneakware did ruin things.
Secondly, when XP mucks up it is pretty good about logging it.
If failure rates are 8%, the administrators must be incompetent. Root cause investigation should bring this number way down. Maybe management wants it this way.
Perhaps the French are smarter, and claim they have to reboot while they get a coffee and a croisant.
Maybe their help desk is trained to to say 'reboot' so they have time to check their fixit scripts, and have time to fake an extra troubleticket to get the numbers up.
The social aspect is interesting: people resent reliable productive platforms, as it gives them fewer means to goof off, or semi plausable reasons for missing a deadline.
What's funny is our IT dept. reccomends that we install the google toolbar for it's pop-up blocking. I simply used Firefox instead, but can not access any intranet sites with it. So for the internal stuff I use IE.
As to the crashing, my XP laptop and Win2K home machines need about a reboot a week on average. My linux boxes and my Win2K "server" (client build) which sees little to no console activity, run for weeks and up without reboots, and all the reboots I've neede to do were because of me.
This leads me to believe that the bulk of Windows is fine and that explorer and the other UI programs are the source of most problems (sans spyware).
-nB
whois gawk date unzip strip find touch finger mount join nice man top fsck grep eject more yes exit umount sleep dump
and the type of people impressed by those brags aren't normally the type of people who you want respect from...
I remember a statistic in a Bill Bryson book saying that 2% of the US's yearly electricity use is from workstations left on over night.
Nothing costs nothing
Windows users obviously have a different expectation of "stable" from Linux users
Very obviously, a post from *one* slashdotter along with your opinion is enough to generalize windows users against linux users. I am just sick of the "windows can't get the same uptime as linux". This is bullshit and everyone knows it, or you are just an ignorant of the Windows thing, so please stop talking about it.
I had a Win2k server in my garage during the last 3 years. It has rebooted 4 times: 3 power outages, one hardware change.
Period. Installed and running during that time:
. Lots of P2P
. Webserver
. Mail server
. CVSnt
. NetTime
. DynDns update service
. Lots of MPEG-2 encoding for overnight encoding (can't sleep with my PC on)
. I probably miss a few
So please, stop trolling next time.
Write boring code, not shiny code!
"Windows users obviously have a different expectation of "stable" from Linux users."
I've been saying this for YEARS!
A Windows user will say "uptime" and mean "time since I had a blue screen" but will NOT count the daily / weekly / whatever reboots they perform.
If Windows starts to go sluggish, they reboot. But they do NOT consider that a break in their "uptime" NOR do they consider that a crash.
# uptime
08:34:13 up 115 days, 18:12, 1 user, load average: 0.10, 0.04, 0.01
That's because I had to move it a few months ago. Everything is current except the latest kernel.
Now I just KNOW I'll see posts from Windows users talking about their "uptime" and so on. But too many of the Windows patches require reboots. Here are the scenarios:
#1. Unpatched Windows box with high uptime.
#2. Patched Windows box with low uptime.
#3. User who does not understand uptime.
4, 8 & 12% of what?
my Windows 2k box at work has been running since (thinks about when the last power outage was) May... so am I to be expecting it to be out of commision now for 15 days really soon?
Thanks to file sharing, I purchase more CDs
Thanks to the RIAA, I buy them used...
I have a Windows 2000 Server and a Windows 2000 Professional machine that I swear to GOD I NEVER have to reboot, unless I'm installing some piece of hardware/software that requires it. I think at one point I had the server up for ~180 days straight, I was amazed at the totals in the "process run-time" in Task Manager.
I sure hope this wasn't on any kind of a network. Last year, Microsoft had 60 (yes, SIXTY) security patches released. That's more than one per week. And yes, each one requires a reboot of the server. Sounds like you and your 180 days are a sitting duck for hackers.
I've got an OpenVMS cluster that hasnt been rebooted since 1999 and an IBM AIX box that's been up since 2001. They just work. Fact is, Microsoft does not have an enterprise-class operating system when every security patch requires a reboot, every device driver install requires a reboot, and every application install requires not only a reboot, but that you close all other running applications during the install. That's just not enterprise level. For these reasons, anyone who uses a Windows server for any sort of 24/7 mission critical application, is just an old-fashioned idiot.
I'd rather be a conservative nutjob than a liberal with no nuts and no job.
Well my Win98 box wouldn't have drug their stats down at all. Of course, the key to that is a clean install where IE and all of it's many components have never touched the HDD. And, of course, rebooting it between once and twice a week. If the uptime gets too long, it definately gets buggy, but it's quite rare for it to actually crash on me - by windows standards of rare at least. I have plenty of exposure to XP, and it doesn't seem any more stable. The figure from this study is congruent with my experience.
Of course the only time I've seen Linux crash was when I had motherboard components failing, and my Mac only locks up about once a year (and almost never gets shut down or rebooted - logging out and back in always seems to be sufficient to fix it when it gets wierd.)
But I have to agree with another posters suggestion - I think that the instability in Windows is primarily from IE, and all the other little things that tie into it. Without that, even Win98 can be reasonably stable (meaning at least as stable as XP, which most people seem to consider 'stable enough' for a desktop or game machine - not stable enough for critical server tasks of course.)
=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Friends don't let friends enable ecmascript.
I don't believe that it takes a lot of electricity. After I had my server running in my house, my summer bill "jumped" from $15.00 to $16.00. I think the monitor is the biggest electricity hog (and since mine was a Linux server, I didn't have one.)
Ok. Here we go. I don't want google for searching. I already have a wonderful tool that lets me to in line searching on the address bar, mostly for google, but it's a lot more flexible. Because of this, I hide the google toolbar, just like I hide the Adobe Acrobat toolbar and the Links toolbar. Unfortunately this has the side effect of stopping the pop-up blocking.
Oddly, once it has blocked a popup from a site for the first time, it seems to continue blocking it even when the toolbar is hidden. This is what gave me the impression that it was just inconsistant.
Like a lot of bugs, it's not really the app, it's not the computer, I'm just not using it like the developer expected.
I worked at a game development company for 5 years and XP was the most stable OS I had ever used in all that time. My reboot percentage on XP is far, far below 12%. I know it's anecdotal, but my experiences contradict their results.
An OS is supposed to MANAGE programs, not be AFFECTED by them.
True unless the program is not segregated from the OS memory footprint (like a driver, or a debugger that allows poking all memory locations). These programs generally require administrator priveledges to install (and setuid patented equivalent to run). Unfortunately, 90% of programs on windows require admin install, and for no good reason. They spew dll's in system directories instead of application local directories, write to HKEY_LOCAL_MACHINE instead of HKEY_CURRENT_USER, or simply use restricted API functions instead of less capable alternatives.
I'm not absolving Microsoft of all of their sins, but most crashes are caused because most windows applications do not follow the microsoft guidelines, and users still install unsafe programs. As an analogy: if all Linux software required kernel.h and required setuid to run, it too could take down your linux box if it crashed. However, most linux programs do not require such priveledges, and if one does, the user generally tries to find an alternative program. So some of the blame should be shouldered by the developers of the applications and the users who install them.
Of course, on the flip side, any linux development guide will tell you that you do *not* put kernel.h in userland programs. Microsoft makes it very developer friendly to do the equivalent (by providing better documentation on all of the naughty stuff I stated above). Well, they've changed on that recently, but they can't erase all of the old articles and books that teach the bad form.
Network Security: It always comes down to a big guy with a gun.
I have two Netware 4.x machines at work that have an uptime of over 800 days each, no kidding.
>>Worms, Adware, Spyware, bad user habits, and just plain crappy commercial software, are all just a bit more than a typical Windows OS installation can handle.
And this makes Windows a piece of crap?
There isn't an OS that can handle all of that. I can write crappy software that will cause problems on almost anything. Worms? They are availiable for just about any (if not all) OS there is. Bad user habits? How can an OS do anything about that? If I want to run as root, is that RedHat's fault? Spyware, keyloggers, network sniffers and all of that is just the user running programs. The OS (ESPECIALLY MICROSOFT) had better NOT tell me what programs I can run on my machine! I want to download and install 6 copies of Gater it had better let me. And it's not the OS's fault if I do. It's not the OS's fault if I do it and don't know what I am doing.
There are many reasons an OS might be a piece of crap. Hard to code for, missing functionality, OS itself have access violations, hard to use, and on and on.
But the reasons you gave just prove the OS is popular, not a piece of crap.