Monday, technology research firm Yankee Group released a report claiming that "Windows 2003 Server is a more reliable server operating system than Linux." Fair enough, I'm open minded enough to give the assessment a fair shake. 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.
Looking into the Yankee Group, and the analyst who contributed to this article, Laura DiDio, it can quickly be decided that they can hardly be seen as an objective source for technology analysis. Yankee has regularly been tasked and paid by Microsoft to provide "objective" reviews for its Get the Facts campaign (see all 184 results from Microsoft's website). The Facts campaign is the same campaign that said one company switched from Linux because they had been effected by the Blaster worm (a Windows worm) on their Linux systems which caused them massive down time and as a result made the switch to Windows Server systems. (read that one for yourself)
Continue reading...
View: Full Editorial
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.
Looking into the Yankee Group, and the analyst who contributed to this article, Laura DiDio, it can quickly be decided that they can hardly be seen as an objective source for technology analysis. Yankee has regularly been tasked and paid by Microsoft to provide "objective" reviews for its Get the Facts campaign (see all 184 results from Microsoft's website). The Facts campaign is the same campaign that said one company switched from Linux because they had been effected by the Blaster worm (a Windows worm) on their Linux systems which caused them massive down time and as a result made the switch to Windows Server systems. (read that one for yourself)
Continue reading...
I'm not jumping to conclusions to say that Microsoft paid the Yankee Group to put out this analysis as they have done in the past. Yankee did make a point of stressing that the survey was not sponsored or supported by any server OS maker. That aside, a research group that has continually been used to produce bias results cannot really be trusted to bring objective opinion to the table, especially when they make claims of being 20% better.
From my experience, Windows servers require far more restarts and downtime than any Linux system. Typically a Linux server only needs a full restart upon upgrade of the kernel, which happens very rarely. Windows Servers will at the least need to be restarted once a month on "patch Tuesday." Are there Windows systems that outperform Linux systems, yes, absolutely. A properly configured Windows server network with redundancy and backup systems will probably perform at the same level (or at least within a few tenths of a percentage point) of efficiency as a similarly configured Linux counter part. If your organization, regardless of if you run Windows or Linux, has 73 days of downtime per year, you need to hire new IT people now!
The Windows Server system is good stuff. My real life company relies on Windows and I use Linux servers for other tasks (such as Neowin) and can say with experience that Microsoft has made great strides in their server platform over the years. However, I refuse to believe the statement that Windows Server provides what the Yankee Group says, and has said in the past.
View: Original Article @ Yahoo News
View: About Laura DiDio @ Wikipedia
View: The Yankee Group | Report
From my experience, Windows servers require far more restarts and downtime than any Linux system. Typically a Linux server only needs a full restart upon upgrade of the kernel, which happens very rarely. Windows Servers will at the least need to be restarted once a month on "patch Tuesday." Are there Windows systems that outperform Linux systems, yes, absolutely. A properly configured Windows server network with redundancy and backup systems will probably perform at the same level (or at least within a few tenths of a percentage point) of efficiency as a similarly configured Linux counter part. If your organization, regardless of if you run Windows or Linux, has 73 days of downtime per year, you need to hire new IT people now!
The Windows Server system is good stuff. My real life company relies on Windows and I use Linux servers for other tasks (such as Neowin) and can say with experience that Microsoft has made great strides in their server platform over the years. However, I refuse to believe the statement that Windows Server provides what the Yankee Group says, and has said in the past.

I wrote about the talk in my blog, it ended kind of funny: http://www.silentblue.net/mtarchives/000946.html
WTF is this doing on our news site? There's already a nice long argument thread about this in the forums; looks like the author's shameless whoring of his own take on the matter, forced down our throats by his mod-power.
As for the "20% more uptime" claim, we run into this sort of thing all the time in statistical annalysis- Odds are, they meant to say "20% less downtime", which is perfectly viable. ("Less down" means "more up", right? Wrong.)
WTF is this doing on our news site? There's already a nice long argument thread about this in the forums; looks like the author's shameless whoring of his own take on the matter, forced down our throats by his mod-power.
It's amusing to what people like you will read into something. How is typing an opinion "whoring"? He's presenting a discrepancy by stating that the "unbiased" source has in fact been hired by MS to do these objective test before.
Have any of you people ever read an actual newspaper? It's that thing they sell at newsstands, made from paper (a pulped tree, bleached and flattened to allow for printing and portability)? There are editorial sections in those too, you know.
If you don't want to read the opinion piece--here's a thought--DON'T READ IT. Skip over. Amazing!
"Coloring" news articles as their posted isn't journalism.
As for the news item this is commenting on, it seems that the common media is misrepresenting the Yankee report, or the Yankee Group has unintentionally misled the media - perhaps a mis-typed news release.
Here is a better source: http://www.iaps.com/2006-Server-Reliability-Survey.html
An example of that claim, please.
An example of that claim, please.
joel
express wont provide any facts to support his claim these fanboys never do
I wonder what such a broad claim serves to make? WHAT kind of servers? I mean, a server can be anything from a home user one being based on a 50 MB large Linux distro running an FTP, to a load balanced multi-computer cluster serving millions of requests per day.
Anyway, Linux is way more common in many server markets, so obviously the OS is doing something right, and stability is waay up on any admin's list, almost always moreso than cost due to economical and PR losses if it's not up. I'll leave it at that and let others draw conclusions.
I think we need more editorials like these.
Last edited by Divide Overflow on 07 Jun 2006 - 17:13
How is pointing out a biased report "fanboyisim"? that's like saying, if Bill Gates said he was god, and somebody pointed out the fact he cant be, then them ending up getting called a fanboy, it makes no sense.
How is pointing out a biased report "fanboyisim"? that's like saying, if Bill Gates said he was god, and somebody pointed out the fact he cant be, then them ending up getting called a fanboy, it makes no sense.
It's clearly marked "Editorial" in a big red circle. If you still want to moan and bitch and whine, then go ahead and stop coming to Neowin. Please. You'll do us all a favor.
It's clearly marked "Editorial" in a big red circle. If you still want to moan and bitch and whine, then go ahead and stop coming to Neowin. Please. You'll do us all a favor.
I don't care if it's marked as an editorial. Anyone posting flamebait like this on the front page on this site is just looking to stir up a controversy. And just a thought. . you're bitching about my whining. . how ironic.
The situation here is the same as bitching because a bag of nuts contains nuts. It's labelled that it's a bag of nuts, and you know it's going to be a bag of nuts when you open it. But you still moan as if you're surprised at it being nuts?
Replace nuts with editorial. Shut up and get on with your life if you don't like nuts.
Edit: And for those of you bitching about an Editorial being on the front page of Neowin, Editorials have been a part of Neowin longer than most of you have and they will continue to be. Feel free to debate the content of them all you like, but if you don't like editorials, don't read them, it's that simple.
Last edited by shanepitman on 07 Jun 2006 - 18:06
I believe I presented my opinion that it is fanboy crap. Someone griping about one thing and praising another does not belong on the frontpage of a news site.
It belongs there if it's marked "EDITORIAL". And it is labelled that way-- in a big red circle for those who have weak eyes. If you don't like editorials, don't read it.
Just because Microsoft asks a third party to do a test does not automatically make the results biased. SOMEBODY has to ask these people to do the tests, right? If the open source community asked, could we automatically assume the results are biased in favor of Linux?
It just seems retarded to think an organization is just gonna waste their time on a project without *some* party expressing interest first. To therefore automatically assume bias makes you look freakin' paranoid.
The moral is, you cannot accept any summarized results without also knowing the details of the testing. In this case, there are barely ANY details.
Does it make me "freakin' paranoid" to doubt the conclusion of this report?
Does it make sense to you that a Windows sysytem has an average of 20% more uptime than a Linux system? Really? Because if Windows is close to 100% uptime (let's just say 5 out of 5 days uptime for a work week), then that means the Linux box was only up for about 4 days, and was down all day Friday. Every week.
Yes. I doubt this report. No sysadmin would accept 80% uptime on Linux or Windows.
The article doesn't state that Microsoft did or didn't give specific parameters, so could you possibly be sure that they did?
Yes. I doubt this report. No sysadmin would accept 80% uptime on Linux or Windows.
That's not actually what it means, it means that if the Windows server was down for say… 20 hours in the 365days it was running the Linux server was down 20% more (24 hours), albeit the article is badly written and doesn’t make this very clear.
That said… this article is still pointless because for all we know they might have configured the servers wrong and the downtime being there own fault.
"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."
Ok, so on average you're going to have, at most, 20 hours of downtime per year. And they did say that Redhat servers were the worst, so we can assume they would hit the 20 hour mark. Therefore, in order for Windows Server 2003 to have 20% more uptime, it would need to have only 16 hours of downtime per year (4 hours better).
UNIX servers technically have 50% more uptime than Redhat, if you look at it the same way this report did. Their main message was that ALL the modern servers are very reliable, and being 20% better is actually not as "OMGAMAZING" as the news poster thought.
Is it a deceptive way to document survey results? Probably. But who wants to hear "Windows Servers stay up 1% longer than Redhat servers per year"?
Of course there are exception of the rule but a big group of linux admin don't care about backup or a alternative plan.
While for a normal condition windows can have a low uptime in comparison with linux, in a danger situation windows can also have a low downtime.
for 365 days to be 20% more than linux uptime, linux would be up almost 305 days/year.
305 x 1.2 = 365 (365 x 0.8 = 292 is not the same, but close 1/1.2=0.8333333)
I agree with linked, they've definitely mis-reported the results with some incorrect math.
Gee, where else have I seen incorrect math today?
What the article probably meant was windows has 20% less downtime, which again is not equivalent to any other statement relating downtime or uptime or differences thereof and involving the number 20% directly.
Remember, we're COMPARING DIFFERENCES, NOT THE ACTUAL NUMBERS.
In fact, you can't determine the percentage difference in uptime based on the percentage difference in down time.
Example1:
Windows uptime: 9000 hours, downtime 1000 hours
redhat uptime 8750 hours, downtime 1250 hours
Windows has 20% less downtime, windows has 2.85% more uptime
Example 2:
Windows uptime: 19000 hours, downtime 2000 hours
redhat uptime 18500 hours, downtime 2500 hours
Windows has 20% less downtime, windows has 2.63% more uptime.
2.85 != 2.63
if you're confused, remember (actual-expected)/expected x100= answer.
If you're comparing windows to linux, linux is the 'expected', but if you're comparing linux to windows, windows is the 'expected'.
If you read details of the actual report you can figure out that isn't REALLY what they mean, but the way that they make it sound that the conclusion readers can jump to.
When oh when will these server wars end??
Commenting has either been disabled on this article or you are not logged in. Click here to login or register, its free!
Note: Anonymous commenting is disabled in order to keep the quality of responses to a high standard.