Secure your Network Advertise on Netcraft About Netcraft Join Netcraft Site Map

What's that site running? Web Server Survey SSL Server Survey Explore web sites News

Netcraft Web Server Survey	Reports  Graphs  Mechanics  Developers  Servers  Most Requested Sites  Archive

Jobs available at Netcraft

The Netcraft Web Server Survey is a survey of Web Server software usage on Internet connected computers. We collect and collate as many hostnames providing an http service as we can find, and systematically poll each one with an HTTP request for the server name.
In the September 2002 survey we received responses from 35,756,436 sites.

iPlanet is the sum of sites running iPlanet-Enterprise, Netscape-Enterprise, Netscape-FastTrack, Netscape-Commerce, Netscape-Communications, Netsite-Commerce & Netsite-Communications.

Microsoft is the sum of sites running Microsoft-Internet-Information-Server, Microsoft-IIS, Microsoft-IIS-W, Microsoft-PWS-95, & Microsoft-PWS.

Platform groupings are here.

Around the Net

Microsoft gains 3% this month as register.com continues to fluctuate between using a Windows and Linux front end. With the register.com domain parking system causing a 3% move in the by hostname count whenever it flips between Linux and Windows [though not making any great impact on active sites], we think presenting a view of the web counting by ip address has merit. This has the effect of boiling off all domain parking, HTTP/1.1 shared hosting, and eliminating duplicate counting of the same server when multiple domain names, such as netcraft.com, netcraft.net and netcraft.co.uk all point to the same site. 

When counting by ip address, the supposed volatility of the past few months disappears, with Apache showing a three percentage point increase from 51% to 54% since the start of 2001, and Microsoft unchanged at 35%.

Finding Application Servers

"Can the Web Server Survey identify application servers?" has been a common question. The strictly correct answer is "no". However, we have noticed that sites running the IBM and Oracle personalised versions of the Apache web server very often are also running application servers from those vendors, and in practice this seems to be a reasonably good indicator.

Plotting the numbers of ip addresses responding using these servers over the year to date shows some interesting things. As expected, IBM seems to have a significant lead over Oracle, but perhaps by less of a margin than expected, given the publicity and attention span generated by Websphere product marketing. There is also a strong regional divergence, with IBM doing much better than Oracle in the US, but this lead being less pronounced in Europe & Asia. 

Solaris sites curiously slow to upgrade

A couple of months ago we highlighted the low numbers of sites migrating to Apache/2.0, and contrasted it with the speed at which site administrators adopted Apache/1.3.26 which contained a fix for a potential buffer overflow problem. 

If anything more surprising is the slow adoption of new versions of Sun's Solaris operating system. Solaris 9, released in May this year, is running on fewer than 1000 web site ip addresses found by the September survey, and there are roughly twice as many sites running Solaris 2 & Solaris 7 as are running Solaris 8, released in March 2000.

Historically, slowness to gather upgrade revenue has usually been a portent of trouble to come for web technology vendors, and the figures coincide with Sun's difficulties generating revenue and profits over the last eighteen months. By contrast, Windows .Net Server, which is not yet scheduled for release, has almost as many as ip addresses as Solaris 9, including some impressive, high volume sites, such as Nasdaq.

Sun would reasonably point out that their boxes typically cost a lot more, and the upgrade cycle for more expensive kit could be expected to be slower.

Reports and Interactive Queries

Fair Use, Copyright

Netcraft Web Server Survey	Reports  Graphs  Mechanics  Developers  Servers  Most Requested Sites  Archive