Analysts Claim Improved Reliability for Linux Servers

Mainstream Linux distributions for servers have caught up substantially with Unix in terms of reliability over the past year, while Windows Server 2003 downtime has risen by nearly 25 percent, according to a Yankee Group survey.

The research firm's survey also noted a significant rise in enterprise interest in Ubuntu, previously known primarily as a desktop operating system.

The 2007-2008 Global Server Operating System Reliability Survey presents a substantially different picture compared to the results of the last such survey in 2006, in which Windows administrators reported less downtime than their counterparts who used Linux — a result that stirred up controversy at the time.

Link: ZDNet

Report a problem with article
Previous Story

Woman's Divorce-By-YouTube Is 'Scary New Step'

Next Story

Windows Vista One Year Vulnerability Report

10 Comments

Commenting is disabled on this article.

"Mainstream Linux distributions for servers have caught up substantially with Unix in terms of reliability over the past year"
Huh? Since when did they have any catching up to do??


"Windows Server 2003 downtime has risen by nearly 25 percent"
Ha ha ha ha... yeah, ok. That would mean Windows Server has less then 75% uptime. Windows 95 has better uptime than that. I think Windows Server 2000 was over 98% uptime and I doubt 2003 is less.


"Windows administrators reported less downtime than their counterparts who used Linux — a result that stirred up controversy at the time."
What's so controversial about that? Windows Server isn't as bad as Linux fans would like to think.


This article makes no sense. I didn't think Linux servers ever had a problem with reliability/uptime. As for their "stats" on Windows Server, three words: Get The Facts.

When they talk about downtime rising 25%, they don't mean that uptime has dropped to 75% uptime. They are talking something like a change from 99.9% uptime (0.1% down) to 99.875% uptime (0.125% down).

I didn't use the actual numbers, but just as an example that was easy to follow. Downtime up 50% doesn't mean uptime down 50%.

And I think that most downtime is for patching. Hardware should evenly affect all commodity type of servers. So patches, prepping for updates/upgrades, and possibly repair for inadequate protection causing a malware spread would all affect the measured reliability of these units.

...Yankee Group said it used intrusion-detection and authentication mechanisms to ensure accuracy.

HUH?
Anyone want to explain that last sentence from the article?

And with all due respects...
The results are based on an online survey of 700 users from 27 countries.

Either I'm completely wrong, or we are not as close as everyone thought we were to running out of IPv4 addresses.

This is like saying that:
"we've surveyed 700 people from 27 states and we can see that Obama is going to be the next president of the U.S.A."

Or maybe...
"700 people in the European Union think that Hitler would've made a great leader"....

You can find 700 people ANYWHERE you want to look... people who will agree with you.

WHATEVER!

(Regression_88 said @ #4)

HUH?
Anyone want to explain that last sentence from the article?

And with all due respects...

Either I'm completely wrong, or we are not as close as everyone thought we were to running out of IPv4 addresses.

This is like saying that:
"we've surveyed 700 people from 27 states and we can see that Obama is going to be the next president of the U.S.A."

Or maybe...
"700 people in the European Union think that Hitler would've made a great leader"....

You can find 700 people ANYWHERE you want to look... people who will agree with you.

WHATEVER!


yeah, the population used in this statistics is much too small to make any kind of conclusion. Still, ubuntu is a great server system, i have been using it for over a year (after using debian for 5 years) and hadn't had any downtime caused by ubuntu or the dell server itself :). On the other hand, my windows server 2003 streaming server has also been running without downtime, so yeah

Well, considering this is a population change that is measured year after year with the same methodology, I would say that the delta is significant.

Can you use this to say that globally, the marketshare of "X" is "Y%"? No. But can you look at this and say that "X" improved by "Y%"? Certainly.

You can also determine a confidence interval based on the sample size of the population you are measuring.

Oh God! My Statistics lectures are all coming back to me now! Aaaaaahhhhh!!!!!!!

Great news. And good to see Ubuntu getting some of the recognition it deserves. It's still a way off being completely mainstream but it is without a doubt the best desktop Linux distribution i've ever used since I started with Linux around 11 years ago (can't believe it's been that long!!)