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 The study was originally made by Acadys and Microcost and gathered data from 1.2M machines belonging to about one thousand companies over a period of one month in seven different countries."

Comments

English Version

[WhatsAProGingrass]

English

babelfish translation with usual mistakes

[Anonymous Coward]

28% of the time devoted to the couple transport/Internet, 2% with Excel

To launch the impression

15/09/2004.

What makes the employees one to their computer? It is with this thorny that question has study undertaken by Microcost - in collaboration with Acadys - sort to answer. Year investigation whose objectifies is not to supervises the users goal who wishes to poses the bases of has reflexion around the rationalization of the costs have glances management of park.

During one month, 1 285 500 working scannés stations were near has thousand of companies distributed in 7 European countries (France, Germany, Switzerland, Belgium, Spain, England, Italy).

First carryforward, has to use spends one average two hours and fifteen minutes per day one its dated-processing station. With time that it devotes for more than one quarter (28%) to Internet/transport couple. The remainder of time, the applications office automation, the trades applications and the Windows to explore respectively occupy 17%, 14% and 9% of the use of year employee. The 17% of the office automation applications station-wagon up into 15% for the 2% and text processing for Excel.

With company thus may find it beneficial any to modify its policy of software licence according to the use in order not to pay has complete office automation continuation principal yew the exploited tool remains the text processing. According to the study, 10 software concentrates 67% of the use. With figure which amounts even to 89% in the industrial sector, whereas it is limited to 42% At the service companies.

In more of the dated relating to the uses of the software, the FRIENDLY software (At the origin of information receuillies for the study) makes it possible to obtain figures have glances reliability of the operating systems Microsoft. Thus, the average failure misses requiring has restarting of the system is measured around 8% per session. This appears fluctuates largely according to the version of Windows. Indeed, Windows 2000 obtains has failure misses of 4% and NT4 of 3% whereas Windows XP flirte with the 12%.

Lastly, the study reveals the use of paid have glances impression. Zero paper is not topicality since 10 pages are printed one average per day and to use. They corresponds to 3 gold 4 orders of impression of which the half are intended for local printers, other half with printers networks. However, yew the cost of year reaches impression has few hundred of euros when it is carried out one has printer network, it is multiplied by five when it is carried out one has local printer, because of the consumable price of the ones.

To also note, without surprised, that 95% of the stations customers are equipped with has Windows environment, version 2000 being prevalent At the professionals. In place under 42% of the stations, this version largely replaced Windows NT 4 which counts nothing any more goal 16%. Have for Windows XP, it breads to find its public, in particular At the industrialists who choose to 83% for Windows 2000. Only the service companies cuts 5% of to their dated-processing park under general Windows XP while the average is around the 2%.

Behind all these figures, the company of council recommends several solutions to the dated-processing directions in order to rationalize to their management of dated-processing park. Among these recalls of good control, the company quotes successively the recourse to the light customer, the uses of software Open source, the optimization of the management of the licences and the increase in the duration of renewal of the material park have well have software.

Re:Perhaps is the user base of those versions?

[Tenareth]

Our entire user base (Over 1000 machines) has been moved from WindowsNT Workstation and Win2k workstation to Windows XP as a global rollout for our company (40,000+ machines). Given the same userbase, and same admins building the machines we have seen XP behave much worse than NT or 2000 ever did.

This is in a completely controlled environment, where we can use GPO to insure extra software is not installed on the machines, etc... unlike the older installed base.

Re:Perhaps is the user base of those versions?

[mrchaotica]

Fine, don't RTFA, but could you consider reading the summary, maybe?

gathered data from 1.2M machines belonging to about one thousand companies

These weren't home users!

Re:Biased

[Coryoth]

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.

I would suggest that my "per session" rate of failures in Linux is quite high. Sure, I don't get kernel panics, but if X locks badly (locking out the keyboard) then my session is pretty much gone. Rebooting X is considerably faster than rebooting the machine.

The real reason that my "per session" rate would be high is that I hardly ever log out. I run a session until something comes out that convinces me to log out (travel, new kernel, or some sort of problem). Sessions last weeks or months.

Jedidiah.

Re:Perhaps is the user base of those versions?

[Coryoth]

came here to say exactly what you said. The amount of clueless people downloading spyware, viruses, and just general crap onto thier computers is ridiculous, and I'm suprised that the failure rate isn't higher. However, if we were to take a look at the professional usage only, where there are IT depts and such supposedly taking care of the machines, I think that the numbers would be drastically reversed.

According to the article there were no home users involved in this. It was all company workstations from about 1000 European companies. That means it pretty much is all in managed environments with an IT dept looking after it.

The best I can find is this (excuse my babelfish translation) from TFA:

"To also note, without surprise, that 95% of the stations customers are equipped with a Windows environment, version 2000 being prevalent at the professionals. In place under 42% of the stations, this version largely replaced Windows NT 4 which counts nothing any more but 16%. As for Windows XP, it pains to find its public, in particular at the industrialists who choose to 83% for Windows 2000. Only the service companies have 5% of their data-processing park under Windows XP while the general average is around the 2%."

Which is about the best I can find for figures breaking down how the different versions were distributed. It seems like XP was largely uncommon except at service companies (and was then still uncommon), so maybe you could claim low sample size - but there were 1.2 million workstations in the total sample, so I don't think that'll wash either.

If someone with far better French than me could provide a proper translation of the relevant paragraph I would be grateful.

Thanks.

Jedidiah

Re:Perhaps is the user base of those versions?

[ThosLives]

The reason I don't shut down isn't because I need to keep running or want to waste power. While it is true that shutting down when I don't use my computer would probably save me some electricity dollars, the startup wear-and-tear on the hard drives and even electrical components is greatly reduced by leaving a system on all the time. Parts tend to fail a lot less frequently if you turn them on and leave them on...it's actually surprising the stress you put on even solid state devices during power-on/power-off transients (you ever notice how light bulbs typically burn out just as you turn them on or turn them off? There's a reason for that... check out what happens to current through a simple R-L circuit during step transitions in voltage.). This concept is true even of light and heavy machinery - it's why jet engines are rated on number of start/stop cycles in addition to hours in operation, and why most large industries don't like to stop and start their plants.

So, I keep my system up as much as I can for reliability, not for "ooh look! X days up without a reboot!" bragging rights.

Re:Biased

[GoMMiX]

Windows average uptime.. listed in days... Linux average uptime... Should I list this in days, months, or years? Seriously, though, I've had Linux servers running and used frequently that were not restarted for upwards of a year or more -- and even then it was because of a flood, power went out - generator was submerged (so obviously shut off), and UPS's ran out of power after 20 mins = everything got rebooted. I have 'never' had to reboot a Linux system because of various parts of the OS or other programs not functioning properly which would be fixed by restarting the machine. I don't care if you're a Windows lover or a Windows hater, everyone knows if something doesn't work right in Windows -- restart, it just might start mysteriously working again. I know of no other OS that behaves this way. Not that I care, really - I like both OS's. I personally perfer to use Linux, but everyone at my company is 50 year old women - when it comes to the thought of training them on how to use Linux... forget that!

Re:Perhaps is the user base of those versions?

[1u3hr]

Many clueless users believe that rebooting fixes many problems

Funny, I believe that. It does fix many problems (such as resources disappearing due to memory leaks or application crashes). Not permanently, they recur, but short of spending a few days reinstallng everything, it's the best solution.

Re:Perhaps is the user base of those versions?

[grub]

It was at least 3 years and at the University of North Carolina according to this page. Search that page for "Server Missing No More".

Unless, of course, there was more than one Novell server walled in at a university for several years... :)

Windows XP vs Linux Driver Support

[Lord Byron II]

I bought my dad a new pc and while I was waiting for it to arrive, I took some of the accessories for his machine that I had bought locally and tried them out on my Linux box. All worked flawlessly, including the usb dialup modem.

Get the new PC, get Windows installed, get the updates, plug the modem in and halfway through the driver install the machine would reboot. Three times I went through this. I tried the Windows native driver, the driver on the disk, and the driver from the manufacturer's website.

Note that the modem came with XP drivers and did not come with Linux ones!

After hearing for years how Linux is always playing catchup in device support, it was a sort of nice surprise to find a device that worked flawlessly on Linux and was beyond hope on XP.

Re:why the switch?

[Tenareth]

The company decided that support would be easier if the entire international company ran a single image, allowing for global rollouts of software more easily. This was partially created by some problems with some major rollouts on a global scale because of different versions of Windows behaving so differently.

Also, Microsoft wanted to use us a proving ground for AD on a global scale... however, the switchover has been so painful that we still aren't fully AD enabled. Issues with major incompatibilities with WindowsXP and our in-house developed applications has been a major stumbling block.

There were also several hardware upgrades we had to do due to the increased requirements of XP over 2000 and NT.

Re:Perhaps is the user base of those versions?

[frp001]

I'm tired of reading Microsoft sponsored research that attempts to make Windows look better than it really is.

As a matter of fact this is not the main subject of the article. The research was carried out by accadys and Microsoft throughout Europe to find out how users used their machines.(The title of the article is about the fact that 28% of user time is spent on messaging/internet -- I wonder if they calculated my time on /.)
Finally the article concludes by saying that Acadys recommends using open source software.

Re:Perhaps is the user base of those versions?

[Phisbut]

If someone with far better French than me could provide a proper translation of the relevant paragraph I would be grateful

Here's a human translation from a French-Canadian

We must also note that, unsurprisingly, 95% of the computers are running on a Windows environment, with version 2000 being the most used among professionals. Win2k, running on 42% of the computers, largely replaced WinNT4, which now only runs on 16%. As for WinXP, it barely found a good public, especially among industrials which prefer Win2k 83% of the time. Only the service companies have 5% of their total computers running WinXP, while the general average is around 2%.

Prime example of why the STORYIES need modding.

[nlinecomputers]

If every a story itself was a troll this one is it. I hate Windows too but the story is misleading as Taco refers to it. It only 8% of windows FAILURES need rebooting as the solution not an 8% failure rate.

I run both Linux and Windows desktops. I reboot about one every two weeks and then usually it is because I've installed a patch or program that requires a reboot to work. In general most of my apps that I run are stable and I get rid of those that aren't.

X-Windows crashes more often for me the MS Windows does. But at least all I have to do for X is restart the X server. MS Windows I do have to reboot. Both are a pain but a full reboot is more painful.

Re:Perhaps is the user base of those versions?

[hopethishelps]

With XP I only reboot once every 2 weeks to once a month. In my personal experience XP is a much more stable environment.

Windows users obviously have a different expectation of "stable" from Linux users. In my office we have just 2 Linux machines but both are heavily used, one for C++ development.

I just ran "uptime" on them. One has been up for 99 days (I remember shutting it down to install a DVD-rom drive about that long ago) and the other has been up for 127 days. Of course I keep them both up-to-date with security patches, but since they're both Debian, that's just a matter of typing apt-get update / apt-get upgrade occasionally. No reboot needed.

Re:Perhaps is the user base of those versions?

[Bull999999]

I had a Win2k server in my garage during the last 3 years. It has rebooted 4 times: 3 power outages, one hardware change.

So you are saying that you never patched that server?

Granted, from personal experience, a well setup Windows machine on good hardware is pretty stable but I believe that the reason why Windows machines cannot acheive ultra long up time is due to the required reboots after certain patches, although 2000 requires less reboots than NT4 and the same thing can be said for 2003 vs 2000.