TheAge reports: The Top five of the top 10 hosting providers for the month of June, measured in terms of those whose sites experienced the fewest failed requests and provided the fastest connection times, are all running the FreeBSD operating system, data from Netcraft shows.
Netcraft: Top Five Reliable/Speedy ISPs Are Running FreeBSD
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSAlert.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Bah. *BSD/*Linux folk have known this truth for a long time.
who are they?
under small, medium and heavy loads makes it a good choice for ISPs / hostings companies. Add to that tight security features and reasonable release cycles and you got a winner.
Here are the results:
http://uptime.netcraft.com/up/reports/performance/Hosters?tn=june_2…
Windows 2000 does good as well, next to Solaris and Linux, down the line.
I’ve just taken a look to the chart, and… Zero downtime with W2K? And on IIS 5.0? Impressive. No, really, I’m not kidding.
Huh. This isn’t a list of “Top 5 Hosting Providers.” This is just a list of 50 hosters (not even top hosters) sorted by failure rate for the month of June.
If you look at the historical data, you see a solaris box breaks the “top 5.”
http://uptime.netcraft.com/up/reports/performance/Hosters?reverse=0
Frankly, the failure rate means little when the numbers are that small. They are more likely caused by network problems than anything else.
Looks like quite a significant number are using windows 2000. Certainly more of the tops hosts using win2k than are using solaris, which could be understandable, and more top hosts using win2k than are using FreeBSD, which is sad.
“Bah. *BSD/*Linux folk have known this truth for a long time.”
I’m not trying to be a troll here, but I’d like to point out from your comment that in the rankings, Linux only scored 1 of the top 10 positions, behind Win2k’s 2.
Really though, fast/reliable hosting provider data really is only a metric for choosing a hosting provider. *BSD and Linux are very reliable in and of themselves, but for a ranking like this, there is just as much, if not more, emphasis on redundancy, infrastructure, support response time, etc.
It should be noted that Yahoo (and maybe others too) are not running vanilla FreeBSD versions, but –at least Y!– have special patches for reliability and mostly, for scalability…
I know that they are a number of Y! engineers taking active part at FreeBSD’s development for example, not sure if they have open sourced back all their changes (they don’t have to, of course, but just wondering if they had).
Very interesting results, but they arent descriptive enough on how the results were obtained. From reading the article, Im going to have to assume that they are doing some sort of active scan certain locations, which isnt going to be acurate to tell you what the site is running, only what its router/load balancer is running. Possibly some sort of passive scan on the results of a full web session to the sites themselves? Ive seen instances where Netcraft had reported a OS of Linux running IIS 5.0… which tells you something, so which result are they baseing this on?
From their FAQ:
Why do you report impossible operating system/server combinations ?
————————————————————
Webservers that operate behind a caching system, load balancer, reverse proxy server or a firewall may sometimes report the operating system of the intermediate machine. Hence reports of ‘Microsoft/IIS on Linux’ may indicate that either the web server is behind a Linux server that is acting as a reverse proxy, or has configured the Akamai caching system such that the first request to the site goes to one of Akamai’s servers [which run Linux], or as in the case of http://www.walmart.com has been configured to send a misleading signature.
Here are the results:
http://uptime.netcraft.com/up/reports/performance/Hosters?tn=june_2…..
Windows 2000 does good as well, next to Solaris and Linux, down the line.
Anybody pass math around here? These statistics mean nothing. Zero.
You don’t know what the architecture is, what the support staff is like, what the networking equipment is, what type of redundancy and clustering each provides, etc. etc.
Maybe the Windows 2000 sites run crappy routers and the FreeBSD sites run rock solid gear. Maybe its the other way around. Maybe one is poor and has their boxes running all the time under extremely heavy loads and maybe the other has servers with lighter loads that give them a better chance of staying up longer.
In short, there is nothing that these statistics prove. Due to the overwhelming superiority displayed by FreeBSD, it appears that perhaps they can make a claim to be the top dog – but even that isn’t provable. For instance it seems odd that FreeBSD caputures the top 5 spots exclusively, but not a single one of the other 15.
Summary: Lies, Damn Lies, Statistics, More Lies, More Damn Lies and some more Statistics.
>> For instance it seems odd that FreeBSD caputures the top 5 spots exclusively, but not a single one of the other 15.
>>
Exactly what I thought! It also seems odd that FreeBSD should overwhelmingly be ahead of solaris in the stability department, especiall as I am presuming that these bigs sites would not be running solaris on intel.
I think these statistics should be treated as reliability of webhosts, not OSes, as they are for many people.
“For instance it seems odd that FreeBSD caputures the top 5 spots exclusively, but not a single one of the other 15. ”
Isn’t 15th place FreeBSD?
“Anybody pass math around here? These statistics mean nothing. Zero.”
I wouldn’t say it means nothing. As you said, it doesn’t mean that one OS is more stable than another. It could just have been a bad month for some companies standardizing on certain OS’s since all the OS’s have 99.999% uptimes. However, it definately makes it clear that Windows 2000 is pretty stable (when compared to NT 4.0), stable enough for big ISPs to use it and stable enough so that they appear in the list. It also shows that FreeBSD is a good free choice, especially companies who want to modify source code without releasing the source. Finally, it shows that Linux, Solaris, FreeBSD, and Windows make good web servers (especially since the big ISP’s didn’t shy away from using them.)
This link might be interesting in this context:
http://geodsoft.com/opinion/server_comp/
Linux only scored 1 of the top 10 positions, behind Win2k’s 2.
Oh yeah? Well at least it didn’t have integrate the web browser into the kernel to get there. ;-p
especially companies who want to modify source code without releasing the source
Well, sometimes some people comment on how the GPL forces companies to do that. If you use GPL software for your business, such as in the case of a web server’s OS, you would never have you release your own modifications. However, if you sold the new modified OS as your own product, then you would be required to give the source code of that product to your customers at no additional cost to comply with the GPL.
This is cute and all, but we’ll see if y’all can keep it up when 2.6 systems start rolling out.
Good job you red little thing =) keep beating the competition and people might start choosing quality rather than hype.
w00t w00t =)
One might wonder if they will start to measure sysadmins skills soon aswell. I think that the only thing this survey proves (and Im not even sure it proves this) is that some Os’s is more secure/stable by default then others. In the longrun it has very much to do with how skilled your sys admin is. I mean if you know how to optimze your system/make it more secure its gonna make a BIG difference in the end compared to if you dont now jack. I think that most sysadmins out there either don’t know jack, or they might not even care about getting their systems 100% optimized and/or 100% secure .
And with Jack I DON’T mean Jack Daniels .LOL
just my 2 cents
This is cute and all, but we’ll see if y’all can keep it up when 2.6 systems start rolling out.
Pardon me, but the endless releases and modifications are why Linux doesn’t show up as much in these surveys… If you recompiled your kernel and rebooted as often as they release patches and updates, people would swear your uptime looked like an NT4 box…
Oh yeah? Well at least it didn’t have integrate the web browser into the kernel to get there. ;-p
Win2k doesn’t have IE integrated into the Kernel.
Pardon me, but the endless releases and modifications are why Linux doesn’t show up as much in these surveys… If you recompiled your kernel and rebooted as often as they release patches and updates, people would swear your uptime looked like an NT4 box…
This is part of the problem, however, the real root of the problem lies in the fact that on average, there are just as many security flaws for Linux distros as there are on Windows, several requiring reboots. How many times has RedHat issued security fixes that affected the Kernel in the past few years? What Linux needs to obtain good uptime is a hot-patchable kernel. However, I would think that this would be extremely difficult.
The BSD world is very different, I’ve seen very few security flaws actually affect any of the BSDs kernels in the past few years.
Granted, all *Nixen have one big advantage over Windows for uptime – the only time you should have to reboot is to update your kernel, since all daemons can be restarted at runtime. Windows admins aren’t so lucky.
Wouldent your connection/bandwidth be more important than your OS?
I for one would never use FreeBSD for my Servers (for a desktop maybe) but my servers will run OpenBSD or Gentoo Linux for as long as possible.
By the way. Whats the longest uptime anyone here has had? And what os did u run at that time?
I for one would never use FreeBSD for my Servers (for a desktop maybe) but my servers will run OpenBSD or Gentoo Linux for as long as possible.
Ummm care to explain why?
FreeBSD’s motto is “the Power to Serve”
It’s the best damn server OS I’ve ever seen, give me just one good reason why FreeBSD shouldn’t be used on a server!!!
Another question…why would you choose Gentoo for a server out of all the Linux distros to pick from?
Perhaps this comment was a joke…i’m hoping it was.
I for one would never use FreeBSD for my Servers (for a desktop maybe) but my servers will run OpenBSD or Gentoo Linux for as long as possible.
You must be kidding. FreeBSD is a great server OS. I use it as a desktop OS, but it’s not really designed as one. Using Gentoo on a production server is not a good idea because it’s such a new distro that it can’t be considered stable. With FreeBSD you can just follow the Stable branch to get bux fixes and security updates.
With FreeBSD you can just follow the Stable branch to get bug fixes and security updates.
Oops, I meant the Release branch. The Stable branch will retrieve new features along with the bug fixes and security updates.
Any server with such a high uptime is probably administered very badly. Every OS has had (security) patches in the kernel from time to time that require a reboot. A box that’s been up longer obviously hasn’t been patched…
My boxes won’t ever have an uptime >6 months, because I update them to the latest OpenBSD release every 6 months
Updates always include a newer kernel (for every OS I know of that is), so updating requires a reboot
Now, do you _really_ want to miss out on all the kernel patches for your OS for the last 3 years just to have a 3 year uptime? If you do, I hope you get rooted soon :p
Of course you can always cheat your uptime, but how lame is that? IMHO, a short uptime (<1 month on average) could be a sign of an unstable system. A long uptime (>1 year) is a sign of a bad sysadmin
I don’t think *BSD is a bad OS, but I definitly don’t think that *BSD is more or less stable than Linux. I bet there *COULD* be servers running for years on a 1.* Linux kernel if not upgraded. Actually upgrading the kernel (or harware failure) is the only reason to reboot *BSD or Linux at all.
Besides, Netcraft is receiving it’s uptime values from the server itself. Uptime is a value which can be manipulated easily on the server. Not the least thing a ISP would do to polish their statistics…
Just to think about.
the most downtime servers?
“Ummm care to explain why?”
More secure by default. But hey im just paranoid.
“It’s the best damn server OS I’ve ever seen, give me just one good reason why FreeBSD shouldn’t be used on a server!!!”
U can use it if you want but I prefer OpenBSD. One could debate cons and pros for the rest of this year but that will just turn into another flame war.
“Another question…why would you choose Gentoo for a server out of all the Linux distros to pick from?”
Have u used Gentoo? I love the ports like system on it and would never go back to any other Linux dist. But hey that’s just my taste.
“Using Gentoo on a production server is not a good idea because it’s such a new distro that it can’t be considered stable”
I have a box that has been up and online for +1 year now and I run lot’s of things on it. I guess u could call that unstable.LOL. But cmon i mean being stable or not has more to do with which kernel you’re runnig.
Also keep in mind that the Apache Project’s web servers run FreeBSD
I suppose most of the top ISPs selected freebsd just for legacy.