Windows Servers Beat Linux Servers

RobbeR49 writes "Windows Server 2003 was recently compared against Linux and Unix variants in a survey by the Yankee Group, with Windows having a higher annual uptime than Linux. Unix was the big winner, however, beating both Windows and Linux in annual uptime. From the article: 'Red Hat Enterprise Linux, and Linux distributions from "niche" open source vendors, are offline more and longer than either Windows or Unix competitors, the survey said. The reason: the scarcity of Linux and open source documentation.' Yankee Group is claiming no bias in the survey as they were not sponsored by any particular OS vendor."

Comments

Same as last year.

[Whiney Mac Fanboy]

Lets look at last years survey being debunked in a business week analysis. ('cause I'm sure not a damn thing's changed since last year's study).

The biggest criticism of the study is this:

Only people running w2k3 AND linux were allowed to respond. Hmmmmmn, so how many MS shops with an evaluation linux server (installed by their clueless MSCE) were included in this "survey"

Yankee group can claim no bias all they like - but I am sick of Laura DiDio fud being posted here (Oh she of 'SCO's claims are justified after looking at the source' fame).

Call this ad-hominem if you like, but if someone pushes a POV year in, year out, you tend to dismiss them.

Re:Same as last year.

[grub]

It was by Laura DiDio. They may as well have had Steve Ballmer make the judgement.

Re:Same as last year.

[semifamous]

Another tech site has an editorial article on this report.

From the editorial:
I administrate both Windows and Linux servers and was interested to see this report. However, reading into the article a bit more makes me question the validity of their assessment.

The Yankee Group states that Windows 2003 Server led Red Hat Enterprise Linux with nearly 20% more annual up time.

I had to do a double take when I saw that. 20% more!? Assume for a moment that you have two servers, one running Windows Server 2003 and one running Red Hat Enterprise Linux 4. Assume that your Windows box ran non-stop, without rebooting (which means you probably are not loading any Microsoft security updates) for 365 days. For your Linux box to have 20% more downtime it'd have to only be up for 292 days. If that is the case, your machine is no longer a server and is nothing more than a space heater.

Re:Same as last year.

[mrchaotica]

Call this ad-hominem if you like, but if someone pushes a POV year in, year out, you tend to dismiss them.

You shouldn't dismiss them just because they're consistent; they could in fact be consistently right (e.g. RMS).

Did you perhaps mean that if someone continues to push a POV after their reasoning has already shown to be flawed once you tend to dismiss them because the situation (and their flawed reasoning) is not likely to have changed?

Re:Same as last year.

[grasshoppa]

Only people running w2k3 AND linux were allowed to respond.

Shame they didn't ask me. While my win2k3 server is up and has been for a while, that's a far cry from saying it's trouble free. More than that, my linux boxes have been up without complaint for far longer AND are more trouble free AND are running apps that don't run on windows.

So, were they to ask me, the headlines might have read something like, "Linux more versatile and trouble free than windows counterpart".

I'll grant you, the win2k3 server is acting in a role that wouldn't be done nearly as well as the linux boxes.

Re:Same as last year.

[Crashmarik]

Documentation for linux is bad. Theres no arguing the point

I just switched a box from fedora core 4 to core 5 and was real pleased nobody had bothered to document the changes to the default install of Apache. I also can't count the times I have looked for things on the LDP or the HOWTO's and found yes this is a very good howto but the distribution is entirely freaking different.

Now I'm not saying microsfts documentation is any better, but they make up for it with consistency in the setup. Pretty much once things are set with M$ they are there. By example, You may not like the registry but its pretty consistent in how it works from win95 to win 2003.

That said once a server is setup and in production why the heck will a lack of documentation bring it down ? I have had Novell servers up for 4+ years at customer sites and they don't even get the docs.

Re:Same as last year.

[MarkLewis]

Your math is wrong. 20% more downtime means 1.2 times as much downtime as the Windows box, not 20% of the year.

So if the Windows box is down for 10 hours per year, the Linux box is down for 12 according to the study.

Re:Same as last year.

[peragrin]

Go back and carefully read the study. Windows doesn't have 20% more uptime, Windows has increased their uptime by 20% while Linux was increased by (insert some random number here)

So if windows servers were available 90% of the time htey have now hit 95% but the linux servers were already at 97-99% uptime so they could only increase by a small margin.

Whenever didio writes you have to learn to read in between the sentences. She throws fud around(finding Linux documentation online, when you could simply call Red Hat and ask???? especially for RHEL 4.)

What she wrote was while techincally true, was so twisted as to be a lie. Notice how she refuses to post hard numbers,or other hard data so you can judge for yourself.

Re:Same as last year.

[Cl1mh4224rd]

That said once a server is setup and in production why the heck will a lack of documentation bring it down ?

It won't bring it down, but it might keep it down.

Re:Same as last year.

[Jon Peterson]

Only people running w2k3 AND linux were allowed to respond. Hmmmmmn, so how many MS shops with an evaluation linux server (installed by their clueless MSCE) were included in this "survey"

Err, that works both ways, doesn't it. Think of all the Linux shops with one little windows server they had to have because some app they needed didn't run on *nix. And IME *nix admins will happily reboot a windows box claiming "it's the only solution" rather than spend 30 minutes actually learning something about how windows works.

*nix admins seem to consider learning windows to be somehow beneath them, while windows admins usually consider learning *nix to be either worthwhile but hard, or simply irrelevant.

Re:Same as last year.

[kthejoker]

The parent strikes exactly at what the GP was complaining about: "poor documentation/changed install" = "free to create your own" in the Linux world. If instead it translated to "improve documentation, have valid and constructive changelogs", how much better off would OSS be?

Re:Same as last year.

[nikoftime]

Windows has increased their uptime by 20% [...] So if windows servers were available 90% of the time htey have now hit 95% [...]

This doesn't seem correct to me - if Windows "increased its uptime by 20%" from an original uptime of 90% then it would have 90% + (.2 * .9) = 108% uptime (or read a different way, 110% uptime). Clearly, you didn't mean either of these. But even if we were to read the statement as "decreased its downtime by 20%" we would still have 10% downtime - (20% of original downtime, or 2%) = 8% downtime. So we'd have 92% uptime, not 95% as you have stated.

Re:Same as last year.

[Illbay]

Isn't comparing Fedora to Red Hat Enterprise inappropriate here?

Fedora is "bleeding edge." Major changes are incoporated from one release to the other, with the time between releases only six or nine months.

RHEL is extremely stable and well-tested, and the time between major releases is long. Therefore, documentation for RHEL will be "true" for a long time.

Not the case with Fedora (I use Fedora, btw).

Re:Same as last year.

[PFI_Optix]

(troll) Silence. Your sensibilities offend the Slashdot drones. (/troll)

I'm a Windows admin. It's what I know, and the only OS I have significant experience with. At my last job, the server with the most uptime was a RHEL3 box that only got rebooted when the ERP database performed its semi-annual crash ritual. Compare that to the four W2k3 boxes that were down about five or six days a year on average for various OS maintenance issues (in Microsoft's defense, we were *doing* a lot more with the Win servers, the Linux server only had one function)

Linux is a hard OS to administer without training. It's not something you can just dive into, and a lot of admins get it shoved on them because upper management decides on a software package that requires it. The result? Downtime because the admin is unfamiliar with Linux and doesn't know where to find the answers. So in that sense, this report is spot-on.

I do question the validity of the data, though. It seems like they picked a sample set that would yeild the results they wanted. A better survey would be to review servers with similar functions, regardless of whether users have both installed. It's no secret that Windows admins have a harder time with Linux and I agree something needs to be done to help them (us) take the plunge with confidence...but this study isn't going to have any impact on anything and was just a waste of someone's money. If they're looking to throw cash away, they should be throwing it at me, not studies.

Re:Same as last year.

[Kadin2048]

Yep, I think you hit the nail on the head.

Saying that one has 20% more or less downtime than the other doesn't say anything about the absolute value of either one's up/down-time. Both of them could be terrible servers or both of them could be pulling four and five nines, we'd never know from that statement alone.

Re:Same as last year.

[sharkey]

Windows doesn't have 20% more uptime, Windows has increased their uptime by 20% while Linux was increased by (insert some random number here)

Well, the article states "Windows 2003 Server, in fact, led the popular Red Hat Enterprise Linux with nearly 20 percent more annual uptime."

That certainly sounds like a claim that Windows has 20 percent more annual uptime than RHEL, expecially since the article doesn't state anywhere that the 20 percent figure was an increase over last year. The only improvement statement made was that "...the major server operating systems all have a 'high degree of reliability,' and have showed marked improvement in the last 3 to 5 years."

Re:Same as last year.

[ketamine-bp]

please, if you'd think that somebody who would complain that installing apache is difficult would be able to write their own installation script, i have nothing to say but to stamp the "ignorant" sign on your head.

be realistic.

Re:Same as last year.

[seek31337]

OBVIOUSLY you know nothing about virtualization. When you run 4 copies of windows on VirtualPC, you can get >100% uptime, per instance, up to, like, 110%.

Re:Same as last year.

[Ryan Amos]

I administer both Linux and Windows servers as well. Windows servers (2003 here, specifically, but the same applies to other versions as well) actually work ok and are probably as stable as Linux as long as you don't change anything meaningful on them. Adding users, changing settings, etc is all ok, but don't you dare install anything on a working Windows server without a full, bootable drive copy or a SAN snapshot. That's where Windows servers lose their reliability in my book.

Blanketly saying "Windows is more/less reliable than Linux!" is flat out wrong (or at the very least, misguided) anyway. What were these machines doing? Were they sitting there just passing packets and not reconfigured once, or are they being constantly tweaked and redeployed? How many people were using them?

Uptime is also usually measured in percentages in the business world. I'm willing to bet the author of this FUD saw "99% uptime for Linux, 99.2% uptime for Windows... That's 20% more!"

Re:Same as last year.

[tomhudson]

10 to 20 hours of downtime a year for a server? That's awful!. Heck, the last place I was at the linux box (Red Hat 9.0) only had 2 downtime incidents in over a year after it was hooked to a UPS - one of those was caused by a 6-hour power outage (the power co was installing new trunk lines, transformers, etc all along the highway as part of an upgrade to the provincial grid), and another by a lightning strike that, again, killed the power longer than the hour of runtime for the UPS.

Of course, AFTER I left, it started going down on a regular basis - but what do you expect when you let a Windows Weenie try to admin a linux box?

Re:Same as last year.

[Herkum01]

The longer uptime is due to time that Windows spends booting up and shutting down. Where as Linux will shutdown quickly and cleanly Windows is still chugging along with prompts say "Are you sure you want to close this program?". This is where it gets its longer up time from.

Re:Same as last year.

[rbochan]

...It's no secret that Windows admins have a harder time with Linux and I agree something needs to be done to help them (us) take the plunge with confidence...

You know what?
There is something that's been done.
You can actually download just about every Linux distro... for free ... and educate yourself instead of bitching and whining "It's too hard! I can't do it!". It doesn't cost you umpteen thousands of dollars to purchase the server OS, just takes an hour or 3 on a decent pipe and blank cd or 3.
Most competant linux admins have done so, even if it's to set up a server situation on some age-old hardware they have lying around to learn how to do things. Also, doing so on older hardware usually forces them to learn how to make the installed server OS more streamlined and efficient so that they can do more with the hardware they have on hand.

I for one, have single-handedly set up a local library with a 500MHz Pentium/256 meg ram that handles all their database and file server needs. I did a testbed on a 233 I had sitting in a closet and had everything down to the tweaked config files ready to go over a month before the project came into fruition. The cost to the library was a 128 meg stick of ram and a 100G hard drive, since I donated my time to them, and they already had the other hardware.
I'd never set up a system exactly like that, and it's worked perfectly for them for the last 6 months, with zero downtime. Took me about 3 days to figure out packages and tweaks for their particular needs. Onsite, it took me about an hour from blank hard drive to them being in full production. They put about $200 into it, and it replaced an aging "server" some salesmen had sold the county for about $3000 that they'd never been able to keep up and running for more than a week on their own.

Can't do Linux? Download it and learn yourself. Anything less is just excuses.

Re:Same as last year.

[Rhett's Dad]

I think the only Linux distributions that should be considered in a study like this are the ones that have corporate support contracts available. Bringing in our experiences with other such distributions (like Fedora) isn't relevant because sensible companies should NOT be allowing any OS into their production corporate infrastructure without proper support available.

If you depend on your savvy hacker to set your production boxes up on Fedora, where is your company going to be when Mr. Savvy moves on? Where's your reliable production support now? Support contracts are insurance against your own employee knowledge base as much as insurance against the server's hardware/software.

Re:Same as last year.

[FireFury03]

That certainly sounds like a claim that Windows has 20 percent more annual uptime than RHEL, expecially since the article doesn't state anywhere that the 20 percent figure was an increase over last year.

The article is rather contradictory because after they say Windows has 20% more uptime than Linux they then say:

On average, individual enterprise Windows, Linux, and Unix servers experienced 3 to 5 failures per server per year in 2005, generating 10 to 19.5 hours of annual downtime for each server.

So, lets assume (for the sake of argument), worst case figures for Linux - 19.5 hours of downtime a year - lets make it 20 hours for ease of calculation. And best case figures for Windows of no downtime.

1 year = 365 days = 8760 hours
So for Linux that's 8760-20 = 8740 hours of uptime per year.

Windows is alledgedly 20% better than this, so we get 8740*1.2 = 10488 Hours of uptime. Which is 437 days.

So to summarise, they've said that Linux gets just over 364 days of uptime per 365 days whilest Windows gets 437 days of uptime per 365 days. I want one of those windows servers that can accumulate well over a year's worth of uptime in a year.

Re:Same as last year.

[FireFury03]

How does the average failure generate 2-4 hours of downtime? In our environment, the number is closer to 10 miutes

They also blame lack of Linux documentation for the downtime. I'm failing to see how that argument works though:
1. Only an idiot admin would take the server down for maintenance until they have the documentation needed to do the job.
2. In my experience, documentation for Linux systems is a lot more readilly accessible than the docs for Windows systems. Yeah, it may not come in a big printed book, but there's plenty of it on the web and searching the web is usually easier than finding the right page in a book. Documentation does of course include searching mailing lists, which will often turn up people having the same problem as yourself, along with the solutions they've found (this sort of thing is not generally published in the vendor's paperbacks).

Obviously the admin needs to be well versed in the OS being used. You can't hire a Windows admin and expect them to immediately know how to fix problems on a Linux system. But the converse applies as well - I'm very experienced with Linux but put me infront of a Windows machine and I wouldn't have a clue where to start.

Blaming increased downtime on the OS when it's simply the fact that you employed people who had unsuitable skills for the job is a big cop-out.

Re:Same as last year.

[FireFury03]

Blanketly saying "Windows is more/less reliable than Linux!" is flat out wrong (or at the very least, misguided) anyway. What were these machines doing? Were they sitting there just passing packets and not reconfigured once, or are they being constantly tweaked and redeployed? How many people were using them?

Also, who was administering them? Did they ahve dedicated Linux admins or were they expecting the office MCSE to handle the machines?

Re:Same as last year.

[FireFury03]

Documentation for linux is bad. Theres no arguing the point

I just switched a box from fedora core 4 to core 5 and was real pleased nobody had bothered to document the changes to the default install of Apache.

Whilest I love Fedora Core, I have many years of Linux experience under my belt. I think it is worth pointing out that Fedora Core is really intended as a testing ground for RedHat and not as an enterprise grade system. If you want things to Just Work and be documented you need to switch to something like RHEL - what you're doing is equivalent to playing with a bleeding edge beta version of Windows and complaining that Microsoft didn't bother to document some brand new feature.

Re:Same as last year.

[a_nonamiss]

Boy, the maths in this post seem to be getting screwed up pretty bad, but I'll put in my 2 cents to see if that sheds any more light on things.

Let's use hours. There are 8760 hours in a typical year. (365 x 24)

Let's say your windows server is down for 30 hours in a particular year. That means it has an uptime of 8730/8760 or 99.66%. Your Linux server has 20% more downtime. That's 36 hours per year. (30 x 1.2) and therefore 99.59% uptime. Is anyone really going to notice a 6 hour per year or 0.07% difference in uptime? (remember, we're not talking specific outages here, just a mathematical statistic - not like "Yeah, if that 6 hours was during our peak time")

Maybe I got that all wrong, but that's how I read the statistic.

Re:Same as last year.

[misleb]

Can't do Linux? Download it and learn yourself. Anything less is just excuses.

That is easy to say, but I'm not so sure it is fair. Just installing a Linux desktop or mock server is nothing compared to actually running an maintaining a production system. Sure, you can learn the basics, but you're not going to be proficient (and get good uptimes) until you've had real world problems/issues to deal with. Especially if the Windows admin is too young to have much DOS/CLI experience.

It helps to have some minor function you can put your first Linux server in. I was fortunate enough to start my Linux "career" at a community college setting in 1997 where uptime was not a high priority. I originally got a Linux box running as a router between a token ring and ethernet networks. And i was able to install a semi-private Linux mail server. I also ran a MUD. Eventually I was confident enough to actually setup a real server... and the rest is history.

-matthew

Re:Same as last year.

[zootm]

Can't do Linux? Download it and learn yourself. Anything less is just excuses.

Easing self-teaching is useful, but it's hardly an excuse for poor or sparse documentation. Self-teaching inherently takes more time (and time is money).

It's easy to learn to do simple things (as in your library example), but where complex business needs are involved, things can get more complex extremely quickly. Documentation and tools help greatly at this point.

The fact that one can teach oneself for free is a great advantage of Linux. If one needs to do so, there's clearly a problem, in particular for business use. This is why supported, documented projects are generally the only ones that are touched by business. Now, I can't speak specifically for the quality of Red Hat et al's support or documentation — my *nix work at my work tends to be on Solaris — but if it's of poor quality, that's definately a huge problem for business use of the technology.

While your suggestion to "download and learn it yourself" is fine for building up experience, but if you're the IT technician lumped with a tight deadline to get a system you're unfamiliar with up and running, lack of documentation could lead to time being wasted, and mistakes being made, simply because there's not enough time to trawl through internet boards and the like looking for the "correct way" to do things.

There is, of course, the argument that *nix boxes should only ever be administered by experienced *nix admins, but (especially in small companies), this is not something that it is practical to ensure.

Re:Same as last year.

[alienn]

Believe or not, by linux can do this also. Just get a multi socket server and run debian woody on it. You'll get uptime*sockets...
I ran this and wonderes how I could have 28days uptime in one week... *g*

Cheers
Alienn

Re:Same as last year.

[timrichardson]

This is easy to explain. The Windows server is installed on Mars. This allows 437 days of uptime and still plenty of time for patch installations.

Re:Same as last year.

[iamacat]

I have no idea where you live, but electric and water outages happen in most apartment complexes perhaps once in a couple of month and bring down toilets, internet connections and nearby ATMs. As for cars, they go down for maintenance for at least a day every six month and nobody complains. If I go to ebay.com or something and it's down, I'll just try again the next day, big deal.

Re:Same as last year.

[Master of Transhuman]

What if you put a Linux server on Venus?

Then Windows is from Mars and Linux is from Venus?

I smell a book franchise here.

Re:Same as last year.

[jackspenn]

It is 20% better then the Linux downtime. So that would be that Windows is down about 16 hours a year using the 20 for Linux in your example.

Couple of points as an RHCE that does both Windows and Linux, I can say that the more I am called to fix Linux machines as an outside consultant the more it pisses me off that each system is configured to the personality of the admin who built it and left, rather then a proven and tested standard. That adds to the amount of time it takes to get a system fixed because there are various smtp, pop and imap servers and various ways to do things that could be the issue with e-mail on a Linux machine, that adds to longer discovery time and in turn longer time to final resolution. Counter that with Exchange 2003 that has published best practices and in most cases one or two ways do do something. This should be common sense to most /.ers. Exchange will not send mail. OK, check the mail queues and the services, look in the event logs. Linux system will not send mail, OK, first figure out what smtp deamon they are using, is it sendmail or not, if not, what. Is it running? Oh, I have two sendmail services, one was installed by RPM, the other was compiled from source, but they previous root user remove those configuration files. And on and on sometimes.

Personal experience has taught me two things:

1). Just because I like Linux, doesn't mean it is perfect. Support issues like undocumented server settings, admins who delete or move the source configs they used in building a package and admins who do things "just to be different" hurt Linux uptime. Also when a company has a Linux server and Windows techs, they will let the Windows techs beat on it like monkeys before calling an outside consultant who costs money; that leads to a large part of that 20%.

2). Your post about 437 days was retarded. No doubt retards everywhere will mod you up as insightful and informative, but that makes your comment no less annoying to people with brains.

- Eric

Re:Same as last year.

[Bert64]

In the case of redhat, you can use the standard mail systems shipped with the OS... Infact, you should never install things manually because then you won't be able to update them using the system package management system.

The same problem can occur with windows, people could be running any one of many mail servers on it, and they won't all be centrally updated.

I have encountered the same problems you describe with multiple systems, a consultant sets up the machine and then leaves, it happens with windows too, but less often, and it's much harder to fix when they've made all kinds of weird registry tweaks, usually the fix is to reinstall, leaving the same problems for someone else in the future.

There really is no excuse for leaving multiple copies of sendmail installed, some from source and some from rpm... But quite often it's necessary to do manual tweaks to any system to make it behave in the way you want... There's also no excuse for not installing your packages through whatever package management system exists, so you can keep track of them and update them more easily.

20% more UPTIME?

[Anonymous Coward]

From the article:

Windows 2003 Server, in fact, led the popular Red Hat Enterprise Linux with nearly 20 percent more annual uptime.

That means that Red Hat Linux has to have at least 1,461 hours of annual downtime, which is 60 days. (This is so that it would then have no more 5,844 hours of annual uptime, in order to allow 20% more of that to fit into one year at 365.25 days.)

I don't think so.

I hate writers who don't understand math.

I'm just not seeing it

[waif69]

I have run both windows servers and linux servers over the last 10 years and my experience is higher uptime with linux servers. Windows machines deal poorly with memory leaking apps and need rebooting for every service pack or required update. I only need to restart specific processes with linux when there is a justified upgrade.

Re:I'm just not seeing it

[MSFanBoi2]

I, on the other hand, see just the opposite.

For years the Linux mantra was that Windows cannot do enterprise, wasn't secure, and on and on... however with a good, well trained administrator behind the console of ANY operating system, it can be made secure, it can do enterprise.

here, because of the "shoot first, ask questions later" attitudes of the Linux support team, the Linux environment (limited to some Web server farms, SMTP servers and a few SAMBA servers), the uptime is around 99.0%. The Windows environment, which is a lot larger, over 1000 servers in total (a mix of 2000 and 2003 but mostly 2003) has a current uptime of 99.95%.
No viruses internally, no spyware/malware internally, inexpensive (compared to what IBM wanted to charge us for Linux support across three years), and reliable.

Yes, sometimes Windows works quite well. For some of us it's cheaper and easier than any Linux distro. People sometimes seem to forget while a linux distro may be free, support for it, from both the admin side, and the overall support at higher levels, is far from it.

Yup, agreed.

[Bazman]

Our Windows 2003 TS servers have a much longer uptime than our Linux servers that are accessed from our lab. Simply because fewer people choose to use the Windows service....

Another 'study' by the Yankee Group...

[TripMaster Monkey]

Why does Slashdot continue to even acknowledge 'studies' performed by the Yankee Group? You think we would have learned our lesson by now...

Hard evidence of collusion may be lacking, but it's still patently obvious that Laura DiDio is a Microsoft shill.

Past experience should be enough to show this, but just in case it's not clear enough yet, here's a snippet of TFA:

But standard Red Hat Enterprise Linux, and Linux distributions from "niche" open source vendors, are offline more and longer than either Windows or Unix competitors, the survey said. The reason: the scarcity of Linux and open source documentation.

Translation: "We don't know how to support Linux, so it's Linux's fault."

Also from TFA:

The Yankee Group made a point of stressing that the survey was not sponsored or supported by any server OS maker.

I'll bet they did...when you turn out such a ridiculously skewed 'study', you pretty much have to make certain everyone knows how 'unbiased' it is.

Re:Another 'study' by the Yankee Group...

[cyclopropene]

articles that were not fragrantly biased against prevailing Slashdot opinion

What exactly does bias smell like? ;)

Re:Another 'study' by the Yankee Group...

[Fulcrum of Evil]

What exactly does bias smell like? ;)

Bullshit.

Defensiveness

[Overly Critical Guy]

We'll see lots of defensiveness over this study in the comments, although if the conclusions were different, it would be cheered. Why not accept it and fix the documentation issue?

Re:Defensiveness

[RevDobbs]

Are you serious?

What documentation comes with Windows? The help system is shit, there is little usefull logging or debugging available in many of their services.

I haven't used a Linux distro in the past several years, but I'm sure that their documentation is worlds better than what comes with Win2k/2003. And I can tell you that OpenBSD's documentation is superb. (Not to mention that the *BSDs trace a direct lineage to AT&T UNIX; does that suddenly make them as good as HP/UX? I have never done 3D modeling on OpenBSD, but if the needed application is available for OpenBSD I would pick it over the comercial unixes any day of the week..)

Re:Defensiveness

[digidave]

The defensiveness comes from the fact that Yankee analyst Laura DiDio repeatedly makes ridiculous claims against Linux. She's the one that said Linux definitely stole SCO's code.

I don't have access to the full report, but I wonder how the "lack of documentation" came into play. Was a certified admin working each system? Did the admin call vendor support for help resolving any of the incidents? Was the particular problem experienced by each server the same? Hardware or software problems? Were all the servers configured in the same role? These differences play a role in how each of the operating systems scored.

Re:Defensiveness

[vidarlo]

We'll see lots of defensiveness over this study in the comments, although if the conclusions were different, it would be cheered. Why not accept it and fix the documentation issue?

Because there are no documentation problems. Do you find an OS with a more well documentet API than Linux? More documentation than Gentoo has? The problem is that they have not studied what I'd dare say is the serious users, they've studied those without in-house competence on Linux.

*NIX-admins are probably more expensive than windows-admins, since there is fewer of them. Those organizations running old UNIX's typically have quite competent admins in-house, and quite different hardware. Windows and linux often runs on off-the-shelf hardware, which I guess explains why UNIX beats both of them

With Linux, the effect is double. A lot of companies have windows admins with some level of degree, but those who know unix, works in the serious business with big unix-machines. Those who adopt Linux have typically not used Unix before I guestimate.

What would be interesting would be to see a study between HP's Windows and Linux servers, since they provide the hardware themself, and should have in-house competance on both OS's.

Compare real things, do not compare different things. Anyone remember Microsoft UK's ad? I think it was along the lines of a x86 off-the-shelf with mssql and win2k compared to a IBM POWER machine. Of course, the ad proved that Linux was more expensive.... This reminds me of that.

Re:Defensiveness

[morgan_greywolf]

What documentation issue?

There are boatloads of documentation available. Ever hear of The Linux Documentation Project? Plus, most distributions offer lots of very good documentation. Why there was a Slashdot story just two days ago about the excellent Ubuntu documentation. There are no fewer than 600 books available about Red Hat distros available for sale on Amazon. Not to mention that Red Hat Enterprise Linux itself includes lots of lots of documentation and most of it is available on the Web gratis. Plus the hundreds of open source apps that include very good documentation with their package. Have you actually read the documentation and free books available on the Samba website? It's darned good!

Any perceived documentation issue is Laura DiDiot's head.

Re:Defensiveness

[Kadin2048]

I think there are other issues as well, namely that it's actually easier in a lot of cases to contribute to a project's code, than it is to its documentation.

If I want to modify the software itself, I can grab the latest version from CVS, make my changes, create a patch, and then submit that patch. Maybe it gets taken into the main tree, maybe it doesn't, but in either case there's a known workflow for contributing to the project.

With documentation, it's not so clear. Let's say I wanted to work on the documentation for my favorite project. What do I do? If there hasn't been anything created during development, it can be a pretty daunting task. I think this is why you see a lot of HOWTOs and FAQs, but no real solid documentation: it's very difficult for somebody, especially someone who doesn't read code, to walk into an OSS project and start writing docs. And even then, it's perceived by many as being a really thankless job, even compared to submitting patches into CVS.

I think the adoption of Wikis in some cases have improved the situation somewhat, because they let somebody go in and make a few changes or update the documentation, without feeling like they have to completely rewrite it (though they can), and it also makes the results widely accessible. However, there aren't that many projects with wikis, at least that I've seen so far.

There are a lot of people out there who are interested in OSS and want to make a contribution, but aren't programmers themselves. But there are a lot of people involved in OSS who seem to not really see a need for documentation, or appreciate the effort that's required in making it really good. As long as that's the dominant philosophy, I think you're going to be stuck with the status quo.

What I'd like to see is an agreement at least in principle that documentation is important, and for Wiki-like CMSes to be maintained alongside CVSes, from planning stages on forwards. I know that's a lot to ask of OSS programmers who are used to just sitting down in EMACS and whipping out some code, but I think the results would be worth the effort. There's a reason why commercial projects sometimes have more people working on documentation and specifications than they do on code, and it's not because they need the extra warm bodies to heat the building. It's because good specifications produce good software (in theory), and good documentation produces happy users.

my Math more reliable than Yankee survey

[yagu]

Another article claiming my OS is better than yours, another article with virtually no information, and the information therein is off-the-scale incomprehensible and inconsistent.

Here's a casual observation: the article says, "

Windows 2003 Server, in fact, led the popular Red Hat Enterprise Linux with nearly 20 percent more annual uptime.

" Later in the article, this:

"..., On average, individual enterprise Windows, Linux, and Unix servers experienced 3 to 5 failures per server per year in 2005, generating 10 to 19.5 hours of annual downtime for each server.

" Let's just say a Linux server has 24 hours of downtime a year (higher than the "survey" says). That leaves 364 days of uptime in a year, 365 days in a leap year.

Implied in the article then, a Windows 2003 server would have to be "up" approximately 20% more to satisfy the "claim". Now, I am not a calendar "expert", but I'm having a difficult time believing that Windows 2003 server is up an average of 364 * 1.2, or 436.8 days a year. If it is, I'm buying.

Also from the article: "..., But standard Red Hat Enterprise Linux, and Linux distributions from "niche" open source vendors, are offline more and longer than either Windows or Unix competitors, the survey said. The reason: the scarcity of Linux and open source documentation...."

First, this is a survey, it hardly points to data that support this survey, in my book a no-no when trying to prove a point. Secondly, assuming there's truthiness in this, my inference from the previous paragraph is, "Red Hat would be a little easier to set up and use if it had better documentation..."

Re:my Math more reliable than Yankee survey

[DragonWriter]

Implied in the article then, a Windows 2003 server would have to be "up" approximately 20% more to satisfy the "claim". Now, I am not a calendar "expert", but I'm having a difficult time believing that Windows 2003 server is up an average of 364 * 1.2, or 436.8 days a year. If it is, I'm buying.

Maybe they are measuring "subjective uptime": it only seems like 436.8 days a year when you are supporting a Windows server?

Re:my Math more reliable than Yankee survey

[ratboy666]

Let me quote the article:

"Windows 2003 Server, in fact, led the popular Red Hat Enterprise Linux with nearly 20 percent more annual uptime."

Which part of the sentence is unclear? 20% MORE ANNUAL UPTIME.

To achieve this claim, what would your numbers be?

Note that it DOESN'T say "20% more downtime". It is very clear: "20% MORE ANNUAL UPTIME". The MINIMUM requirement to achieve this is 60 downtime days on the RHEL box.

Note that we ARE being "relative": the 60 downtime days is the MINIMUM. Assuming 100% uptime on the Windows Server. The downtime can only grow.

Conclusions?

Maybe the Yankee Group has some math or writing (or both) problems. Implies that they shouldn't be in the publishing statistics business, which needs both skills.

Maybe the RHEL hardware is very defective. Implies that the Yankee Group is not TECHNICAL enough to do this report.

Maybe RHEL is defective. With that much downtime, I would have called Redhat. Maybe the Yankee Group has a horribly misconfigured system, and is trying to sort it out themselvs. Implies that they are not smart enough to do this report.

Every time: the Yankee Group doesn't have the skills needed. The report should be ignored.

Ratboy

Downtime?

[Alioth]

Three to five down events per year totaling 10 to 19 hours of downtime per year? I'm not SuperAdmin, but NONE of my servers are ever down for that long or that often. Who are they letting run these boxes? What are they doing? Taking the machine into single user mode and recompiling the kernel before rebooting them or something?

Re:Downtime?

[normal_guy]

I don't know. My experience with MySQL, perhaps the most commonly-used application behind Apache, shows a surprising lack of robustness with regards to unexpected power outages or hardware errors. The hours can add up when you're rebuilding large indexes or fixing corrupt tables.

Documentation for running a server?

[nuggz]

How does documentation affect the uptime of a server?

You need documentation to make changes, not to leave the server alone.

If you're making changes you're not measuring the reliability of the OS/software, you're measuring software and admin performance.

Re:Documentation for running a server?

[gowen]

How does documentation affect the uptime of a server?

I keep trying to plug my USB key into the power switch.

BSD

[Fireflymantis]

So where would *BSD fall in. Along with Linux because of the clueless people rebooting it because they don't understand /etc/init.d, or along with UNIX because (I'm linux user myself) BSD users actually do seem a bit on the more experianced side of the fence.

Uptime vs. downtime

[Martin Blank]

Is it 20% more uptime? Or is it 20% less downtime? There's a very, very big difference there -- two months of downtime is pretty severe, and if you have that, you have some serious problems. From the reverse perspective, three nines of uptime allows for nearly nine hours of downtime per year. If that downtime is reduced by 20%, that's nice, but not really noticeable for most users.

More informative version of the article

[neonprimetime]

A more informative Summary of the 2006 Survey

Total Bullshit

[mabu]

First and foremost, the whole nature of the design of Unix/Linux provides a means by which software systems can be updated without any service outage. You cannot do this with any version of Windows. Most Windows-based patches and upgrades require a system reboot, which is downtime. Most unix-based upgrades merely require a quick stop/start/HUP of the services. If their main claim is that updating system components is the basis for downtime, they're smoking crack. Maybe their methodology for testing involved taking the entire system down while they upgraded? Unix doesn't require such drastic measures - Windows probably does, as you probably can't update a running service. By design, Windows is exponentially more prone to downtime in the process of patches and upgrades. It's virtually impossible for them to compare the two OSes on this issue and not be dramatically manipulating the test methods to create bogus results that are in no way reflective of how sysadmins patch and manage their server resources. I call BULLSHIT.

I have unix servers right now with uptime measured in YEARS. There are no Windows boxes that can make that claim. Period. I've had outages on occasion due to DDOS or system probes that caused a process to terminate over the years, but I've never had any type of wholesale outage that you'd typically get with most Windows installations. Does anyone have any details on the methodology of the testing? It's obviously bogus.

Re:Total Bullshit

[metamatic]

I'm not sure if you're going to believe me, but I have a Windows 2000 Server running for about 2.5 years now, serving as a small ASP-driven web server.

So you're saying you haven't installed a service patch to your Windows 2003 box that required a reboot in 2.5 years? Care to post the web server address? I'm betting you won't dare.

WxP Pro

[robpoe]

We have a WinXP Pro box that's been up over a year ...

Another box that's Win2k pro that's been up almost 2...

The one app they run is heavily used (dispatch for a 911 center).

Re:WxP Pro

[Anonymous Coward]

If your windows box has been up for 1 and almost 2 years, respectively, it means that they haven't had security updates applied (which require a reboot). And if your 911 center doesn't keep it's servers patched, you should all be fired.

Re:WxP Pro

[Mancat]

Why not?

Systems like these used in call centers often:

1) Have no route to the internet.
2) Have both external storage drives and USB ports disabled.
3) Do not allow users to log in with administrative accounts.
4) Have proper group policy restrictions in place.

More often then not, even without the latest patches from Microsoft, machines in this state are perfectly secure and stable. Argue if you'd like, but there are plenty of offices I've worked in where the Windows machines aren't even up to SP2, and because they are properly locked down, they're solid as a rock and still running.

Re:WxP Pro

[robpoe]

Oh, you have NO idea of what you're talking about.

"Back in the day" (TM), 911 dispatch was an old green screen with a serial connection to Ma Bell for the ANI/ALI information. Radios were 20 year old cards in a rack of radio equipment. Stuff gets hard to find replacements for, it gets upgraded.

Enter new systems:

The phone switch? Windows controls the user accounts. The phones are Windows interfaces to hardware. Controls which line gets switched to the dispatchers headset. Completely out of my control on those boxes. They're vendor provided, vendor controlled.

The other box? The 2k one? Radio interface control. Controls what channel the dispatcher's headset talks on.

None of these boxes are on the Internet. Security? Uhhm. Ok. These boxes have no external drives, run non-admin user, the phones are on a hub network with each other but nothing else (not connected in or outbound to the Internet), and are not on our main network. The radio computers aren't even networked. The computers are locked behind two wooden doors, inside an underground secure facility that is monitored 24x7 with people who have guns.

Now, as far as the dispatch software? The thing that keeps track of calls? That runs on Netware (soon though to be converted to MS-SQL). That app runs on Win2k or WinXP (vendor suggests NOT running SP2). Those computers DO have outbound internet connections to about 15 hand selected sites (all state or government agencies, or public service sites, like the hospital tracking system, crime victim notification sites). They do have email installed, but the email goes through a server running MailScanner, set to disallow every kind of executable file (and vbs, etc, etc). They are NOT configured to run any Windows Media, no IM. They're locked down tight.

We've had no virus problems here .. for 3 years now. Malware on some computers? Occasionally, but our antivirus detects most Java dropper programs and kills them.

Should I be fired? Doubt it. ;)

Yankee

[Elektroschock]

http://www.businessweek.com/the_thread/techbeat/ar chives/2005/04/the_truth_about_1.html
http://www.computerworld.com/softwaretopics/os/lin ux/story/0,10801,82070,00.html
Laura DiDio, an analyst at The Yankee Group in Boston, said she was shown two or three samples of the allegedly copied Linux code, and it appeared to her that the sections were a "copy and paste" match of the SCO Unix code that she was shown in comparison.
DiDio and the other analysts were able to view the code only under a nondisclosure agreement, ... "The courts are going to ultimately have to prove this, but based on what I'm seeing ... I think there is a basis that SCO has a credible case," DiDio said. "This is not a nuisance case."

Watch the "expert" Laura Didio on video from a credible source:
http://www.microsoft.com/windowsserversystem/facts /videos/didio_video.wvx

Enjoy her!

*lol*

Re:Yankee

[catdevnull]

Uh...yeah...no bias there. Not a bit. Nope.

Looks like she's been getting quite a few free lunches from Microsoft.

Doesn't jibe with reality

[Ritz_Just_Ritz]

How come I never get any of these "impartial surveys"? I have racks and racks of RHEL Linux servers that I only reboot when:

a. a machine suffers a hardware failure (fairly rare) or
b. there's a kernel update that impacts security

In the case of (b), I apply the updated rpms and reboot which normally results in a downtime of approximately 60 seconds for that server. This might happen a few times a year (single digits).

For our small number of Windows 2003 server boxes, it seems that each "windows update" cycle recommends a restart. We'll call that a once a month reboot when Microsoft gets around to releasing their monthly cleanup. Total server downtime is maybe 2-3 minutes (windows takes a bit longer to reboot on the identical hardware used with our Linux machines).

So while I *could* say that our windows servers are down XYZ percent more than our Linux servers, in terms of actual downtime, both platforms are about the same, with Linux seemingly holding a small edge in my experience.

Cheers,

Re:Doesn't jibe with reality

[WillAffleckUW]

I agree, we run our entire lab on Linux, with the exception of my WinXP machine and a laptop downstairs.

Reliability - thy name is Linux - we can't afford to keep rebooting our servers when our medical genetics perl scripts run days and weeks before crunching the DNA sequences and family inheritance statistical inferences.

Maybe shops that are only open for 9-9 each day can, but we have to be up all the time.

Windows Server is nice.....

[ShyGuy91284]

I don't know about uptime, but I used to be a Linux-Only person when it came to servers. After recently falling into a job where I have had to administer Windows servers, I'll admit they are slick...... I picked up workiing with them a hell of a lot easier then I would have a Linux server (if I was new to it). Good LAN support features, ISA, Exchange, license management, fairly easy remote user/computer maintenance..... I'm probably going to give it a shot for my next home server once I get the parts. Although the software is costly if you want to learn it as a hobby (I'm getting it for my home server through MSDNAA).

Re:Windows Server is nice.....

[pl1ght]

I'm glad you admitted it first. I work with about 300 Linux(slackware, redhat flavor) webservers. Which are great, no problems for the most part, but we also run about 45 Win2k/2k3 IIS/MSSQL/AD/Exchange servers for our Intranet apps and i must also admit that we have no problems with these guys. They are very very easy to manage and setup and ive yet to have a "crashing" problem that wasnt hardware related on either OS. To each his own. Both have their strengths.

What?

[C_Kode]

"Red Hat Enterprise Linux, and Linux distributions from "niche" open source vendors, are offline more and longer than either Windows or Unix competitors, the survey said. The reason: the scarcity of Linux and open source documentation."

Hmm, thats odd. Linux documentation has always been in great abundance. It's getting information about how OS interals worked that caused me the biggest OS to Application head-aches. (Both Unix and Windows)

On a broader note, said Yankee analyst Laura DiDio

Ohhhhhh, I see. Laura DiDio had her nasty little Microsoft-lead hand in this survey.

Math Nitpick

[colinrichardday]

That would be about 304 days, as 20% of 304 is 60.8 (304+60.8=364.8). The 20% must be taken as 20% of the RedHat uptime, not the Windows.

But yeah, that's way too low for RedHat.

Re:Math Nitpick

[Coryoth]

More to the point -- is "uptime" the opposite of "downtime" or is it "uptime" as in the output of the "uptime" command? With the latter, the 20% difference is at least plausible;

Ah, now we get to the heart of the matter. Obviously Microsoft has managed to pull ahead by padding the output of the uptime command: 20% more characters means 20% more uptime!

Re:Math Nitpick

[Ucklak]

So basically you're saying it's crappy Microsoft programming again.

If you have a Win2K3 server and a Linux server side by side and they've been running for 120 hours as measured by an independent timepiece,
Linux uptime would report

14:28:27 up 5 days, 0:0, 1 user, load average: 0.00, 0.17, 0.66

Windows uptime would report

Current System Uptime: 6 day(s), 0 hour(s), 0 minute(s), 0 second(s)

Re:Math Nitpick

[knifey]

Well, the Windows machine would report 432000 seconds. Which is a much bigger number than 5 days. Maybe the people doing the report just have maths issues, as is already suggested by the 20% more uptime.

PS, as someone who administers both Win and Linux servers, I gotta say the report is so full of sh!t it's scary. 233MHz half dead Fedora C3 machine has about a 99.95% uptime. Win2K3 machine with latest hardware, ~99.2%. Um, lemme think about this.

Yankee group website uses win 2000

[olddoc]

According to Netcraft, they have a whopping 4 days since last reboot: http://uptime.netcraft.com/up/graph?site=www.yanke egroup.com/ They also go with the bulletproof reliability of MS IIs

Re:Same as last year = more BS

[moeinvt]

What the hell kind of shops/businesses/people are they surveying? People that have their servers running for a couple of days a year??

"According to the Yankee Group's annual server reliability survey . . . Windows 2003 Server, in fact, led the popular Red Hat Enterprise Linux with nearly 20 percent more annual uptime."

I would think that most businesses want to have their servers up 24/7/365 minus a few hours of scheduled reboots and upgrades, and unless something breaks or crashes. So, assume a Windows 2003 server had PERFECT uptime record for the year.

365/1.2 = 304.17. So, in order for Windows to beat Linux with 20% more uptime, they're trying to say that a server running RHEL is down more than SIXTY DAYS a year? My BS meter just crashed.

The thing is, it SOUNDS plausible.

[Theovon]

See, I know far too little about system administration. If I were to try to run a Linux server without help, it would be down all the time. If _I_ wanted a server, I'd pay someone a service feel to maintain it for me, and it would be up all the time.

So, it seems to me that ON AVERAGE, Linux servers would be down more than others, because so many people would be trying to admin themselves. The lack of documentation would definitely be a problem. (Actually, there's plenty of documentation. FINDING it is the problem. I don't know enough to come up with the right Google search terms! And posting to usenet is hit or miss.)

The question is what the uptime is like for Linux distros where you're paying out the ass for support (like you would for Windows or UNIX anyway). That's got to be such a small portion of Linux servers that it's not dragging the percentages up.

The real metric should be UPTIME / ($$ spent on support).

Be careful about those divides by zero.

Clear proof!

[Anonymous Coward]

See for yourself:
http://uptime.netcraft.com/up/graph?site=www.micro soft.com
http://uptime.netcraft.com/up/graph?site=www.linux .com

Obligatory Debian post.

[khasim]

I just switched a box from fedora core 4 to core 5 and was real pleased nobody had bothered to document the changes to the default install of Apache. I also can't count the times I have looked for things on the LDP or the HOWTO's and found yes this is a very good howto but the distribution is entirely freaking different.

100% agreement. Which is why I prefer Debian (although I'm migrating to Ubuntu).

I can easily clone a production server and walk it through the upgrade process ... over and over and over and over ... and submit bug reports for any and all problems. All during the "beta" phase of the next distribution. I did that prior to migrating my servers to Sarge last year.

apt-get dist-upgrade

It is truly awesome. You can test and re-test the entire process every time they release a bug fix for any of the packages you'll be using. (Yeah, you can do it with gentoo, also.)

If it is a doc issue....

[bblazer]

I am not surprised. Documentation of many open source projects (including linux) is often very poorly written and/or not maintained. Being a good code writer does not necessarily translate into being a good documentation writer. Major software companies hire whole teams of doc writers, and the results are (many times) much better than those that come with OS projects. This has been one of my fundamental points in the never ending discussion of things that are hindering wide spread adoption of OS solutions.

My own study

[onebuttonmouse]

Debian Sarge x86: 63 days, 19:43
Debian Sarge PPC: 61 days, 12 min
OS X 10.4 PPC: 51 days, 1:02
NetBSD m68k: 107 days, 37 mins

So, if you want the highest uptime, use NetBSD on a 25MHz 68040. Further, I contend that my study is at least as believable as the article cited in the submission.

Just a few reasons I don't trust this

[martinultima]

I happen to be a Linux developer myself, and other than my own inherent bias towards my own system, I'd have to say that there are a few good reasons I never believe this junk:

Uptime isn't everything, especially as far as security patches are concerned. Sure, the site can run for years on end, but does it have the latest security fixes, especially for low-level kernel and other system stuff that often requires a reboot to finish?

The distribution does matter, at least in my own opinion. Red Hat might be one of the big names, but it's only one of many distributions – and a lot of them, especially ones based on Slackware Linux such as my own, tend to be considerably more stable and reliable (I may be wrong here, but I've used both systems, so I have at least some experience). So don't blame every Linux system just for Red Hat's problems.

History repeats itself, as a lot of previous posts have shown – they always make these extraordinary claims about Windows vs. Linux, but it's always the same people, and they're always proved wrong in the end. Since everyone else has already made this point, I'm just going to leave this one alone, and continue to the next:

UPTIME ISN'T EVERYTHING! I've already said it once, but uptime is only one of many different statistics as far as this type of thing's concerned. Sure, it's important that servers stay running as long as possible, but honestly, there are other important factors to consider as well – I'd rather have a Linux system that's occassionally offline than a Windows one that's always on, because (a) the Linux system would likely do what I need it to with far less work, at least in my opinion; (b) the Linux system would likely be much more secure against outside threats than the Windows one; and (c) face it, I'm biased, and so's everyone else who does this type of thing.

Just remember, there are three types of lies: Lies, damn lies, and statistics. I don't think we really need to say anything else. Q.E.D.

MS studies are not just FUD

[spyrochaete]

I wrote a Microsoft-funded white paper last year with the assistance of two subject matter experts - a Microsoft expert and a linux expert, both certified veterans of their fields. The goal was to compare the processes required to set up and administer various services in Windows 2003 Enterprise vs. Red Hat's and SuSE's boxed enterprise server NOSes. Because the white paper was intended for internal use only, we had 100% control over what services would be tested, how to evaluate them, and how to present our findings. We didn't evaluate uptime per se, but I feel my comments are relevant since installation and maintenance contribute to server and client downtime, ergo, uptime.

We compared many factors including user management, authentication, "ghosting" new machines remotely, remote application installs, file sharing, delegating authority to subordinate administrators, and much much more. The Windows and Linux guys would work on a "lab" side by side, often peeking over to see how the other was doing. At the end of each lab we'd all have a discussion about the number of steps, any problems, company and community support, the ease/frustration factor, and how it went overall. We wrote about all these factors and rated them on 10-point scales per lab, and condensed those into one comprehensive graph showing overall ease-of-use of each NOS.

Long story short, Windows came out on top by a huge margin in every field - ease, usability, intuitiveness, support, everything. In fact, the only topic where Linux came even close to Windows was in community support, and even that was only 50% of Windows' score. At the end of the project the Linux expert garnered a lot of respect for Windows and quashed most of his prejudices. Needless to say, MS soon compiled our white paper into marketing materials and stuck them on http://www.microsoft.com/getthefacts (but it's been replaced by more recent studies).

I was a little disappointed that we couldn't expand the scope of the test to put stuff like Apache and Squid and mySQL through the paces, but the topic was enterprise administration, not publishing live services. I also would have liked to have tested custom installs of other linux flavours like Debian or Slackware, but neither product had a specific enterprise distribution.

So don't be too quick to label all pro-Windows studies BS or FUD or other ignorant catch-all acronyms. I personally was funded by MS to spearhead an impartial study, and MS management had a genuine interest in improving their products. I can't speak for the study in TFA, but my own was conducted with nothing but integrity and truthfulness.

Re:MS studies are not just FUD

[waveclaw]

Like every other system administrator I have to write and read reports or run tests on hardware and software. To shortcut a lot of problems I start by critizising the (far too often flawed) methodology of any study I get before I base a decision upon it. This is not ment as a personal attack, but (maybe because of marketing mangling) I saw real flaws and a lot of bias in the case study that was originally used in Get The Facts. The biases I claim to have seen were subtle and very nasty, but of a completely different nature than the one in TFA[1].

I wrote a Microsoft-funded white paper last year with the assistance of two subject matter experts - a Microsoft expert and a linux expert, both certified veterans of their fields.

Case studies are an important part of understanding a wide variety of phenomena, however, in textbooks containing them there is often a disclaimer: those were particular people, with particular skillsets in a particular situation[2]. Neither I nor anyone else (say Microsoft's marketing department) is justified in generalizing that situation to anyone else. Hence the demand for surveys such as this one. You can translate the metrics used in the Get-the-facts paper into variables and then show that many others, with very different situations still show these results. Unfortunately, this article does no such thing. There is no specification of what kinds of servers, the platform configurations or even the application loads.

We compared many factors including user management, authentication, "ghosting" new machines remotely, remote application installs, file sharing, delegating authority to subordinate administrators, and much much more. ...
We wrote about all these factors and rated them on 10-point scales per lab, and condensed those into one comprehensive graph showing overall ease-of-use of each NOS.

I would hope that, given an expert on any topic that I'd get a good ease-of-use for that topic. At that level of operator skill and performance, which I have tried to mention is very atypical, I would surprised if the huge resources of Microsoft had put out a failure. Was there was something that the Microsoft product could do the Linux one could not[3]? What features were missing? Why was that feature missing? That was then, this is now, how do those compare today? The pace of change in Linux features is not determined by a single vendor[4].

Long story short, Windows came out on top by a huge margin in every field - ease, usability, intuitiveness, support, everything.

The reason systems administators exist is because of their skills at doing things that are not easy. Otherwise they don't keep their jobs very long (but this is the same for any job.) I really can't argue for or against ease as a metric.

I would hope that with the huge desktop penetration that Microsoft's OS leads in intuitiveness. Now if your Windows admin had grown up in a Macintosh home, used a Mac and home and on his workstation at work I'd be inclined to consider the intuitiveness argument. 20 years ago, that Linux admin would probably have come from a Unix desktop and Unix workstation and Unix or Mainframe server envrionment. How can we be sure that 20 years from now it will be Linux or OS XXX on the desktop? (On the other hand, the byzantine way OSS is developed does encourage only-developer-friendly interfaces.)

MS soon compiled our white paper into marketing materials and stuck them on http://www.microsoft.com/getthefacts (but it's been replaced by more recent studies).

I belive that Novell, one of those 'niche players' in the Linux world (11% Linux webserver share vs 49% RedHat Netcraft 2004,) released a much better take on those marketing materials with it's Get the Truth campaign.

I personally was funded by MS to spearhead an impartial study, and MS management had a genui

Re:They cannot beat my uptime.

[Rich0]

Uh, I can't remember the last time I had to patch my linux kernel due to a security issue. Maybe there was an occasion in the last year, but that is unlikely. That is the only thing in linux that requires a reboot.

As far as the other stuff you mention goes - none of that requires substantial downtime. Sure, if you're making an application change you might need to work out what dependencies need to be updated, but it isn't like you're going to do that while production is down.

If you're running a server that you care enough about to bother reading uptime surveys, then you're not going to make any changes to it until you've performed the change in an identical test environment following a documented procedure, and have tested the change to ensure the system works. Then you just follow the same procedure again in production and it will work there as well. And running a different OS isn't going to make a difference in your application update plans, except possibly for stuff like package-manager utilities or other tools to manage your apps. On linux you'd actually have more support for this stuff - if you need to have good control over your production server with some app not supported by the OS vendor then you're better off making your own RPM for it.

If you generally upgrade applications by fiddling with the production server until it works, then you're really not the target audience for server uptime reports. That kind of stuff will kill uptime on any server.

Linux Documentation issue? For MCSEs?

[Skapare]

It's probably more a case of MCSEs that don't grok the concepts of Linux and how it is documented. The survey was supposedly limited to just shops that run both Windows and Linux. That means you are likely dealing with a bunch of MCSEs that have been working with Windows for over a decade and have only in the past couple years been given Linux to also administer. If such a survey were limited to shops that had been running both systems for an equal period of time and have people on staff who are specialists in each system and have equivalent levels of experience (for example the Windows admin has 10 years Windows experience and the Linux admin has 10 years Linux experience and both have been working at this particular shop for 4 years, which has been running both Windows and Linux for the past 6 years), then I think it might be able to show the true differences and similarities. But I don't think this is anything Ms. Didio is capable of doing.

Uptime

[TripHammer]

I don't give a rip about uptime per se. What I care about is being able to reboot on my terms..and sadly...Windows usualy sets those terms.

Really a non-issue

[MarkLewis]

I know that zealots from both sides will indulge in the opportunity for some bashing and grandstanding, which is fun and I enjoy a good smack-down as much as the next Slashdotter, but I just wanted to inject a little reality check. This study doesn't mean anything about OS quality. Numbers always lie, and even if you're not trying to make them lie, they're sometimes useless anyway, as in this case.

Linux has a different user base than Windows and UNIX. So the fact that 20% of the responses to the survey show higher Windows uptime doesn't mean that for the same usage patterns Windows is higher quality. To show that, you'd need to compare only sites with very narrowly targeted usage patterns which differ (as much as possible) only by their choice of operating system. This study didn't even attempt that, so making any claim whatsoever about the relative quality of operating systems based on this data is fallacious.

Instead, this study CAN shed some insight on the type of people running different OS's, and their type of usage.

Since UNIX had the highest uptimes, you might conjecture that conservative people run UNIX. Or you might guess that since UNIX market share is currently eroding in favor of Linux and Windows servers, UNIX servers are more heavily weighted towards older, established systems that aren't in early development stages as much as the up-and-coming OS's.

Conversely, since Linux had the lowest uptime, you might guess that Linux has a higher percentage of fresh new applications running on it.

(Note that I'm not saying that there's conclusive evidence of these guesses here. I'm just saying that when considered together with other data, these are the kinds of conclusions you'd be able to draw from this sort of statistic).

flawed logic

[rs232]

"Linux distributions .. are offline more [because of] the scarcity of Linux and open source documentation"

Even assuming that were true how does a lack of paper cause an OS to bork?
And how does a lack of paper cause the OS to stay offline longer?
Is it because it is sulking?

How was the data collect the data?
What was the methodology used?
Where are the figures?
What criteria was used in selecting the participents?

How is a marketing firm or Ms. DiDio for that matter, suitably qualified to comment on OS security? Finally are these people really the unbiased commentators they claim?

"Windows servers recover 30% faster from security attacks than Linux servers" Laura DiDio July 2005

"Indemnification is a serious, potentially costly issue for enterprises" - Laura DiDio Oct 2004

"For the time being, she said, Linux has an apparent advantage simply by virtue of a lower level of connectivity" - Laura DiDio April 2004

"hype notwithstanding, Linux's technical merits while first-rate, are equivalent but not superior to Unix and Windows Server 2003," - Laura DiDio Mar 2004

"This has the potential to turn into a twentieth century witch hunt," "There is a visceral anti-Microsoft sentiment in Europe." - Laura DiDio Sep 2003

"The entire Linux community is saying to customers, 'You're on your own,'" .. "That's not a place I want to be." - Laura DiDio Aug 2003

True, and I think eventually OSS docs will shine

[blueZ3]

Google is where we all go for answers these days.

I think it's interesting that MS sells a lot of software (Office, for example) that has less-than-great documentation, and that this is also a complaint people have with OSS. Currently there's a market for commercial documentation for both types of software--my local Borders has lots of books on both Linux and Windows Server. But eventually I expect that OSS documentation will improve to the point where it's better than what is provided in proprietary software, since there are people both willing and able to contribute to the documentation effort in OSS.

True story: I was once offered a full-time job where the owner of the company essentially told me he was looking to hire a Microsoft Office expert to provide office suite "consulting" to his staff. I didn't take it (as a technical writer, I want to actually write) but it said a lot about the state of usability and documentation for Office.

Re:They cannot beat my uptime.

[segfaultcoredump]

This is the postfix gateway, external web server, dns server, etc for our little (1000 employee) company:

root[loki:/]# w
  10:57am up 1030 day(s), 21:27, 1 user, load average: 0.05, 0.02, 0.04

This happens to be a Solaris 9 system. It has never crashed. Actually, over the past 5 years we have had 1 software related bug take down one of our solaris systems (multipathing bug in the FC drivers when used with active/passive disk arrays). This is based on an environment with 40+ solaris based servers (running a wide variety of services, this is not a '40 identical servers shop')

The best our windows boxes can manage is 6 months (and that is if we skip a few of the security patches).

I can guarantee that during the past 3 years, every single one of our windows systems (60+ servers) has had an issue that is core OS software related (not counting the security related ones). Kernel memory leaks are the most popular (file server reboot every 115 days or it will freeze up). Security worms are another fun one, but kinda rare today compared to the good old days.

Nothing against "Communication Majors" really.

[golodh]

Really ... I wouldn't say a word against Communications Majors (as miss Didio is according to the Wikipedia), except where you propose to rely on them to tell you anything accurate, or anything about technical matters.

Others have already commented on the lack of clarity, the need to read between the lines, the absence of the most elementary numbers and facts about this "study" (as in: how many respondents, how recruited, how many rejected and why, how was uptime defined and measured, what were the uptime numbers, (contingency table by OS this year, contingency table by OS previous year)).

If any students read this, let me take this opportunity to warn you. Submit a "report" like this to any serious faculty and look forward to an F grade. Unless you're a "Communications Major" obviously, in which case you'll be complimented on the flow of your prose.).

I'm guessing here of course, but I think that the real study was conducted and written by someone totally different, and miss Didio got the write the "teaser": i.e. the part that you can release without divulging any real information that you would otherwise be required to pay for.

Raise your hand...

[PhYrE2k2]

Raise your hand if you have read any documentation included with any software you purchased in the past five years. Anyone? Anyone?

Okay then- raise your hand if you know that there are 600-odd page gorilla Linux reference books out there which may provide documentation should you need it that will be 100x better than anything included with the software.

Raise your hand if you know where to seek help, such as #linuxhelp and #linux on EFNet.

Case in point. Why not put a properly run linux server against a properly run Windows server- that is what it comes down to. A trained, professional, and experienced admin who has learnt the software they are running and know it well, in a specific purpose. Put Linux as a fileserver against Windows as a fileserver with any optimizations possible and equivalent configurations that are agreed upon beforehand. Put Linux versus Windows as a Web server with a knowledgable admin. This `good at neither` system doesn`t work!
-M

Re:They cannot beat my uptime.

[plague3106]

Because individually those bugs might not allow remote root, but in clever combination they would? Or perhaps there's also a hidden issue which wasn't widely known that the patch also fixes?

Also, you don't need to reboot Windows as much; often times you are told to because of a file lock, but you could easily unlock the file by stopping services which are using it at the time.

There are very very few servers in the world that need 100% uptime; if your server is that critical, why isn't there a fallback server you can rely on to bring the orginal down?

There really isn't any excuse not to patch, unless you want to leave yourself open.

I notice you never answered the question though; perhaps there is a security patch which you've not applied?

Re:They cannot beat my uptime.

[Kadin2048]

I'd probably have uptime like that on an old home server I have sitting in a closet somewhere, unfortunately it reboots every time the power goes out and the UPS drains.

There ought to be some kind of metric for "software uptime," i.e. the delta between the uptime of your actual services (HTTP, SSH, whatever) and the uptime of the building's mains power / network connectivity, etc. I'm pretty sure quite a few servers I've worked on would be at 100%, or close to it.

Otherwise, any time I start seeing surveys like this, I start to wonder about how they've tried to normalize the variations in hardware setups, connection reliability, power service reliability, etc. I could probably get more uptime if I had a bigger UPS and paid for a better internet line, but it's not that important to me; I suspect even datacenter users eventually have to make a determination of how much they want to spend for that last "9" to the far right of the decimal place, and not everyone's decisions are going to be the same.

Re:They cannot beat my uptime.

[marcello_dl]

In debian based distro land, you upgrade everything on the fly. You need not reboot. You need not disconnecting from an ssh session to update ssh. Only security problems in the kernel need a reboot... so linux potential uptime = time since latest kernel hole require a reboot, and actual uptime IMHO is 99% or more of that.

In acronym land, you call BS on TFA, BTW.

Windows documentation

[owlstead]

Yes, Windows is better documented... That is, if you are looking for really shallow documentation. For both Linux and Windows, you are way better off by buying a few good books. The GUI documentation of Linux is pretty worryingly bad, but if you go deeper, it gets better. With Windows, it's just the other way around. Even MSDN is pretty bad and (maybe more importantly) one sided. And, if you are trying to watch it on the machine you are working on, prepare for a reboot; MSDN requires the latest Internet Explorer most of the time. I do not expect .NET to improve this situation, with Java application servers you can just unzip the stuff in a folder and run (as with the VM).

Wrong assumptions.

[alexfromspace]

If you carefully read the quote:

Red Hat Enterprise Linux, and Linux distributions from "niche" open source vendors, are offline more and longer than either Windows or Unix competitors, the survey said.

It does not sound right. Every IT professinal I know (including myself) whose company runs both Windows and Linux agrees that Windows breaks all the time while Linux does not. But this statement is not based on the same assumptions as those under which most people operate under. Why this is so becomes apparent when you read the next part of the statement:

The reason: the scarcity of Linux and open source documentation.

It is apparent that the author considers any server designated to run Linux but not yet installed (whether partially or at all), to be a downtime server. In other words, this study can easily include hundreds of unused machines that should or could have been running Linux. This is completely laughable because only people on crack could possibly agree that a system that has not yet been setup causes downtime. This assumption would not agree with the definition of the word "downtime".

I generally find that whenever Linux is being attacked, it is only through a model with serious logical fallacies that are carefully covered over by seemingly innocent mistakes. In reality these are carefully engineered FUDs designed to sound valid to most common people but failing under any serious scrutiny.

I can conclude from these quotes that the author may feel that Window's point and click interface should somehow justify its inefficiencies compared to Linux. However, Linux's lack of point-and-click gui tools is very old news that got washed away several years ago when tools like Mandrake's free setup tools for Red Hat and SuSE's YAST came about. And besides, it is better to have to learn to setup systems using text config files and then have it run problem free for a year, than to point and click for a day and end up with a system that needs constant attention just to be kept running.

Re:They cannot beat my uptime.

[gonk]

Red Hat released a purely security related kernel update on 2006/05/24:

      https://rhn.redhat.com/errata/RHSA-2006-0493.html

I would be very surprised if your kernel did not have known security issues that you are unaware of. Whether or not the various security issues apply to your environment is another question.

robert

Agreed

[phorm]

For the record, I've been using linux on servers and for my desktop for years.

Documentation for big projects (apache, squid, etc) is usually easy to find. However, when you start running between versions and other isssues, suddenly the waters become a bit murky. Google is often friendly, but lately I've been lucky to find docs in english let along for the version(s) of software I'm using.

I've also been taking my LPI (my employer's idea). It's a freaking linux certification/exam and has no official documentation, other than a general overview of topics. That's right, no course materials, nothing. Even as an experienced linux user/admin I generally don't memorize the dozen different ways to do something, and then find that the one I didn't know is on the exam (which you can't study beforehand, because no documentation).

Sites like TLDP et al are very useful, but a more comprehensive set of documentation (and more up-to-date documentation coming with the software packages) would certainly be a useful thing.

Huh? Wha?

[QAPete]

It always amazes me that the FUD train never seems to get derailed. Much of that I blame squarely on the left-wing nutbag Linux purists (or Open/FreeBSD purists, for that matter) for their crappy documentation holier-than-thou attitudes when it comes to ENGAGING OTHERS and getting them excited about Linux/BSD. Some of it I blame on RedHat's remarkable inability to market their quite excellent RHEL Enterprise Linux. Much of the rest, IMHO, goes to people who suckle on MS's cash-engorged breast....

.... however, I digress. My personal experience here as a Director of IT running HP-UX, RHEL, Slackware and a variety of Windows servers is that Linux (firewall, fileserver, web & ftp servers) run neck and neck with our HP-UX boxes (ERP system) for uptime. My latest Nagios printout shows zero downtime for both this year, and only a .001% difference last year (meaningless).

On the Windows side of things, Win2k3 server performs the best, most outages due to forced reboots from patching/hot fixing. Win2k servers follow closely behind. Both are about .04% behind the Linux and HP-UX boxes for uptime, and much further behind when it comes to the CPU running at 100% for an extended period of time doing whatever Windows servers decide to do overnight (Nagios classified as 'unavailable').

This 'study' is utter dreck. It's flawed from the get-go, and people have to be careful believing what they read. If you are an IT professional, I highly recommend you speak with peers who have some serious experience with Linux before you proceed with any deployment/changeover/rollout. Depending on what your needs are, you might be pleasantly surprised!

A technical note

[cecom]

On Windows it is impossible to delete or replace a file which is in use (e.g. a shared library). The same applies for directories. Thus for any meaningful upgrade you need to restart the applications and often the OS _before_ you can do anything with their files. There are complicated mechanisms for keeping track of files that need to be deleted/replaced after a reboot. It appears that recently they have added yet another even more complicated feature to avoid reboots: http://www.eweek.com/article2/0,1895,1895276,00.as p

Such complicated techniques for a basic thing like an upgrade make me very nervous. What happens if something goes wrong with the extensive bookkeeping in the middle of the upgrade ?

Comparisons and secrecy and independence

[UnknowingFool]

There are always some study that says one OS is better than another. Most often the study is funded by one of the OS groups. That doesn't it necessarily make it useless. What makes them useless is when the details of the study are not released.

These studies present themselves as scientific but they are not. In true science, the data and the methodologies are presented for scrutiny. There could be issues with either or both that would harm the results. True science involves skepticism.

Remember a few years ago when some cult claimed that they cloned a human baby. The first reaction was "Can we see and test the baby's DNA?" When the answer was no, the majority of scientists dismissed their claims outright. The minority reserved judgement until there was actual proof.

Until I can look at the study, I'm not going to believe it. Since no one paid for the study, the Yankee Group does not have any restrictions unless they mean to profit by selling the study.

No Use

[wakejagr]

There is no use. With no real information, this study is crap. It is just throwing more FUD on the pile. One of my favorite bits: I love that Linux is refered to as a less mature operating system. "Yankee Group determined a significant portion of this outage time is attributed to the scarcity of Linux and open source documentation compared to the more mature, established operating systems."

In some ways (support for 3D graphics HW, sound), Linux is not as developed as Windows or MAC, mostly due to proprietary vs open driver issues. In many other ways (portability, support for older/slower HW, virtualization, load sharing across machines, security, customization), Linux has far greater maturity.

UNIX is a totally different issue, but the Linux vs UNIX comparison is moot as far as I am concerned. For the most part, the kernels are the only fixed-in-stone aspects of these OS's. Some things don't exist in the Linux kernel, others don't exist in proprietary UNIX kernels. Choose your poison based on what you need.

In the end, the list of features unavailable in Linux is short and inconsequential when compared to the list of features unavailable in Windows or Sun/AIX/etc. OpenMosix, Xen, and User Mode Linux alone should be enough to overwhelm the Linux downside of making sure you buy a video card from a manufacturer who isn't an ass.

Just to round out my arguement: the other measure of maturity is time-based. Windows NT (follow-on from ideas developed in other versions of windows) was first released in July 1993. Linux (follow-on from ideas developed in UNIX and minix) was first released in August 1991.

ridiculous assertions

[sloanster]

The assertions are ridiculous on the face of it, obviously prepared by someone with an agenda, and not even a bit subtle.

As an IT professional, I can tell you that if any of our linux servers were to go down, there would be people screaming bloody murder all over the place within a few moments. Downtime is unacceptable for infrastructure services, and linux has performed flawlessly for the fortune 100 company where I am employed.

I think as other posters have noted, the key piece of information that was unwittingly leaked, was that the survey was only open to windoze shops, and most likely included some mcse's linux test boxes in the downtime data figues. That's really the only thing that makes sense, as downtime simply wouldn't be tolerated in a normal production environment.

Anyone who is works with linux professionally and is aware of the fact that it's been running 24x7 for years at amazon.com and other firms such as my own employer, will find it quite odd to read about all this extended downtime and the nonsensical reasons given for it.

Linux Sys admins actually update their machines.

[johnnnyboy]

This is just pure FUD.

Uptime is something I see as completely meaningless, the longer the uptime just means the longer the machine wasn't rebooted that's all. Why would a sys admin reboot? After a major update or a new kernel or even after serious patches.
In the world of linux, there's constant improvements, enhancements and security patches in the kernel and in the server software. I think the gap is more to do with Linux Sys admins being more active in maintaining their boxes compared to the windows sys admins.

If windows 2K/2003 servers do have longer uptimes than that to me is just scary, it means those sys admins aren't applying security patches regularly.

My 2 cents.

In my shop.....

[fatboy]

I have to reboot Windows2K3 jsut about everytime an update is avaliable from Microsoft. I started using the system only a few months ago.
I have not had a reboot of the Linux system we use here in well over a year, (448 days to be exact) even though I have updrad applications and applied many patches.

BSD Not Evaluated?

[BanjoBob]

According to various articles scattered around the net, the Unix flavors included Solaris, HP-UX, etc. But, I have seen no references to NetBSD or FreeBSD as a Unix that was evaluated.

While boxes are boxes and OSs are OSs, the application that the server is running needs to be factored in. There are many cases where a BSD server may be a better choice than Linux or Windows just as there are cases where Linux or Windows may be the better choice. I found it interesting that I can find no reference to a BSD Unix in any of the links to the study.

So, since this study has so many unanswered questions relating to function, measurement criteria (what is considered downtime?), application, hardware, etc., the survey is pretty much worthless.

Box+OS is a tool and I use the right tool for the job. One size does not fit all solutions.

What you want is "deborphan" and "debfoster".

[khasim]

Probably better in mst cases to do a fresh install, though. At least you'll get the opportunity to weed out the redundancies in your files.

Ah, I can see that you haven't experienced the Love of Debian yet.

With Debian, grab deborphan and debfoster and you can weed out un-needed packages quickly and easily.

"deborphan" compares the dependencies of each package so you can see packages that are installed that nothing else needs. Delete the ones that you don't need.

"debfoster" shows what all the dependencies are for a particular app. For example, Apache can have all kinds of packages it is dependent upon. If you want to get rid of that app, you can also quickly purge all the packages that were installed as dependencies for that app.

Once you've got the machine stripped down to the basics, just check all the files in the non-home/non-data/non-log directories to make sure that they each belong to a package. Or that you know why you put them there.

It runs sweet.
It runs clean.
It runs exactly what you want.
Nothing more/nothing less.

Which makes patching the box soooooooooo much easier. And it means that you have fewer potential security holes because you're running fewer apps.

Bad examples for a bad result.

[Oriumpor]

As far as TFM it's qualifications draw my suspision. Did they include "devices" running linux as well or just full blown rigs? I can tell you *nix based appliances (unless they're really bad) have very few problems, and don't typically require the constant reboots for system updates that drives down your 99.99..999999 uptime.

Whatever happened to limiting exploitable processes? Windows method of protecting the services is all based around their firewall. Ever try and configure a windows box to run slimmed down? It's a pain in the ass. How about hardened? Good luck, apply the NIST standard lockdown SecPol to a 2k3 box and you'll see what I mean.

Take a *BSD/Trustix(+SELINUX)/Debian(+SELINUX) box install with 3 services AND a firewall in a 100meg footprint, and call it a day. Windows can't compete with the kinda uptime you get out of a stripped down OS. Oh they try with XP-Embedded and the likes but it's certainly not within the same realm of ease to create and deploy the OS that the *nixes give you. Not to mention, how many times have you had to troubleshoot a problem in Windows that ended up being caused by some unrelated service? I can tell you from my experience, it doesn't happen very often on a machine running single digit numbers of services.

On top of which they nicely avoided shops smart enough not to run Windows devices in their nocs, who probably have much better trained staff on the unix hardware and would throw their numbers with nearly 0 downtime figures. How many untrained people new to unix reboot when they could have just restarted a service? etc. This whole thing smells fishy.

Press Release and Interpretation

[makemineagrande]

Here is the note I sent to Laura DiDio - and their PR manager:

You probably should not read the DiDio-bashing going on over at Slashdot today, but I do see what I believe is an error in the presentation of the data in the press release http://www.yankeegroup.com/public/news_releases/ne ws_release_detail.jsp?ID=PressReleases/news.server reliabilitysurvey.DiDio.htm.

The specific statement, "with nearly 20% more annual uptime" is I believe factually not supported by your numbers. Do you mean that Windows has 20% LESS DOWNTIME than RHEL?

"on average, individual corporate Linux, Windows and Unix servers experience three to five failures per server per year, resulting in 10.0 to 19.5 hours of annual downtime for each server."

If RHEL had 19.5 hours of downtime, and WIndows had 15 hours of downtime, this would be 20% less downtime. 5 hours less downtime per year is actually real data and would be useful to the press release.

On the other hand, 20% more annual uptime would actually result in RHEL being down nearly 61 DAYS per year assuming Windows is up 100.000%.Note: 60.8333 days = 365 - (365/1.2)

----------- The report may be correct. The press release is most certainly in error.

Too bad TFA doesn't say 20% more downtime

[Mateo_LeFou]

"Windows 2003 Server, in fact, led the popular Red Hat Enterprise Linux with nearly 20 percent more annual uptime."