Netcraft Claims Apache Now Runs 2/3rds Of The Web

Mr Bill writes "According to NetCraft the Apache web server now owns over 2/3rds of the web. The jump of 2.8% since last month is mostly due to a number of large domain parking sites switching back to Apache from IIS. 'During 2001 and the first half of 2002 several companies hosting very large numbers of hostnames including Webjump, Namezero, Homestead, register.com and Network Solutions migrated to Microsoft-IIS. Subsequently these businesses have either failed, significantly changed their business model, or reverted to their previous platform, and Microsoft-IIS share is now in line with its long term pre-summer 2001 level of around 20%.' See the full report here."

Comments

Not necessarily a good measurement

[taliver]

Numbers that are much harder to get but would be significantly more valuable would be the fraction of web traffic handled by the type of server. Just because I have a hosting company that has 3 sites doesn't mean that I'm getting traffic in the same amount that some other individuals. And MS(make that M$ so I don't get modded down) would tell you that there servers are deployed on the larger installations, the ones that need to higher performance.

(And, I'd expect that if we looked at a graph of traffic, you'd see the GWS getting a significant share.)

NCSA

[chill]

Netcraft really needs to drop the NCSA line on the charts that don't stretch back before 2000.

The only thing that straight orange line at 0 does is give the Sun ONE guys something to point and laugh at. And it looks like they need it.

Re:good

[Anonymous Coward]

It might help a bit, but not a lot. Web servers that belong to a domain, say www.slashdot.com, are counted here, but when you have millions of home machines worldwide still running an open web service on windows, that can overwhelm the statistics.

66% of 'real' websites may be apache driven, but when it comes to viral infection, Joe Normal's home windows box on his cable connection counts just as much an infectable web server as the business down the road that runs a real .com

Define 2/3rds of the Web

[winkydink]

I offer that 2/3rds of all web servers and 2/3rds of the Web are far from the same thing. While I have no firm idea how to accurately measure the Web, I'd offer that either total content or total content that is actually viewed would make for a far more intersting statistic.

Whether this makes Apache's percentage larger or smaller, I have no idea there either. I think that the claim as written is inaccurate.

Re:good

[nmg196]

But IIS usage is NOT going down though! The netcraft graph is a graph of relative usage of each system and adds up to 100%. If you look at the bottom the linked page at the second graph, you can see that IIS usage hasn't decreased at all - it's just that Apache usage has gone up quite a bit recently (ie, there are more total servers tested by Netcraft).

Re:OpenSSL...

[Przepla]

Indeed, but:

However, relying on version numbers to determine the number of vulnerable OpenSSL sites is flawed because vendors backport security patches. So a site using OpenSSL on a Red Hat 9 system will likely report itself as OpenSSL 0.9.7a even though it isn't vulnerable to any of the issues mentioned and the situation is similar for SuSE, Debian, Mandrake, and most of the Linux distributions. Additionally, many of the vendor distributions of Apache have recently started supressing all the extra module information by default, so newer distributions (ones that are not vulnerable) are less likely to be listed.

I'd just add, that FreeBSD does the same thing.

Re:good

[Bert64]

Apache is not really a monoculture atall, not compared to IIS... If you encounter a machine running IIS you can pretty much guarantee it`s running on an x86 machine running windows, it might, but this is a 1/1000000 chance or something, be running on windows on an alpha, mips or ppc... but this isnt possible for any version above 5.0
However, with Apache, it could be running on any one of many OS`s, Linux, FreeBSD, Solaris, HPUX, AIX for instance, and on many different hardware architectures.
This is a good reason for promoting systems such as FreeBSD, OSX, and the other risc systems... If the entire world standardises on x86/linux for their webservers, especially a single distribution, then it would be no better than a windows monoculture.

All that shows is apathy

[GregWebb]

Just because many don't complain doesn't mean they're not being disadvantaged. I could steal 10 pence a day from you and you probably wouldn't notice. Does that mean my theft would be permissible?

Impartial, informed observers have been saying for a very long time that Microsoft are a monopoly and illegally maintain this. That a major customer of theirs (HP, I believe) felt strongly enough that they disliked dealing with Microsoft sufficiently to go on record as stating that if they had alternative suppliers, they would deal with them instead, is surely a strong indication of Microsoft's nature. As is Microsoft feeling able to pressure IBM into dropping OS/2 and later SmartSuite through preferential pricing on Windows. Surely if there existed a sufficiently realistic competitive market in computer software, such tactics would have merely driven up sales of OS/2? It's not like it wasn't getting good reviews at the time.

Microsoft are a monopoly in the legal sense, and there can be no doubt that they have significantly abused this to the detriment of both consumers and the industry as a whole to anyone who followed the trial. That users are too apathetic and uninformed to understand they have lost out is not a defence against the monopoly charge, merely and indictment of the popular media and Microsoft's few remaining competitors.

Re:That's Just Crazy

[lone_marauder]

If anything, apache is closer to a monopoly (though both IIS and apache are far from it)

Monopoly != popularity. Monopoly is taking market share by force rather than by normal market behavior. If Apache had extensions that didn't work right for any other browser besides, say, Mozilla, you might have something.

Please turn the next page in your pamphlet and post accordingly.

Well, that's interesting but...

[carlmenezes]

Wouldn't a breakup by a measure of the size in bytes of content served by the various web servers make a much more realistic figure?

I mean, if the traffic logs and stats are not available for all the sites around, surely, a measure of the size of the content would give one a fair idea of where the heavy weights really lie?