logical.order.of.chaos

If you look at this netcraft.com domain name registration graph, you will see a clear exponential pattern. 

Total Sites Across All Domains August 1995 - November 2007 

There are two lines here.  In blue you have the number of domain names registered.  This is the total number of domains people are snapping up.  In red you see the number of active domains.  These are the ones that are actually doing something with the domains.  It is a much smaller number. 

The blue line, or the domain registration curve clearly follows an exponential curve, and is doubling all the time.  Can this continue?  You might think that this trend has to top out at some point, won’t we run out of usable domains?  The answer is, not for a very long time.  Every year new extensions come online which starts another gold rush each time.  Companies are always trying to tie up new domain names. 

Some of the domain name purchases are simply tying up their brands on each new extension.  Some of this traffic comes from domain cyber squatters.  This in itself is an interesting topic.  There is a type of domain purchase that is being used right now, that are technically legal, but a bit questionable.  Domain Kiting is a technique that works within the domain registrar rules to try a domain before they buy it.  Kiting allows the buyer to float a domain and never reall own it.  This can be used to purchase the domain for longer periods of time to test it for free.  They simply keep registering and dropping domains within the 5 day grace period.  This allows them to see if the domain is good or not without actually ever purchasing the domain.  Domain Tasting is the generic term for when someone buys a domain and sees how much traffic is generated.  If the domain doesn’t generate enough traffic within the first 5 days, the domain is dropped and a new domain is purchased to test.  It is not evident from the report how much of this domain tasting is causing the exponential curve.

 The red line here indicates the number of active domains.  These are domains that are actually up and running.  This excludes parked sites, so that means it is only sites that are actually running content.  It is not clear, based on where the curve sits at this point, if the active sites curve will follow the same exponential growth as the domain registration curve, or if it will follow the flatter, cubic curve.  I will go out on a limb and predict cubic growth.  The reason for this prediction is simple.  Companies tend to lock up as many domains as they can, but only use a small portion of domains they actually own.  I think in a years time this curve will show the pattern it is moving towards.

 The next graph on the page is very interesting.  I noticed this a while back.  I thought I commented on  it, but cannot find the post. 

Market Share for Top Servers Across All Domains August 1995 - November 2007

The blue line here is the market share for Apache web servers, and the red line is the market share for Microsoft web servers.  It is difficult to tell what is exactly causing this, however I think it shows a clear trend.  Apache is falling in popularity, as Microsoft IIS is gaining.  The biggest gains for IIS are companies like Myspace that are using Microsoft technology.  The gains from Myspace are some of the biggest gains in this report.

I think there are a number of factors here.  The biggest factor is that in 2003 Microsoft released a new, more stable Server OS, with a far more stable version of IIS.  IIS6 was the most stable and well thought out version of IIS yet.  A few features, like the low attack footprint in the default settings makes IIS6 the most secure version of IIS.  Out of the box you have far less open and available services.  What this means to the administrator is, by default very few services work, and must be turned on explicitly.  This is good news because, many attacks have been perpetrated over services that are open which the administrator doesn’t even need open.  One big example of this was port 445 or the Universal Plug-n-Play port.  This port was open and active by default on 2000, but closed on 2003.  2000 server computers were attacked by a virus a few years back that wreaked havoc on 2000 servers that were not protected by firewall.  As usual it takes years for a new OS to be adopted fully by business.  You can see here where this may have been the cause of the increase once server admins realized that this was a more stable, and easier server to use.  The change in direction happened in about February 2006. 

Another factor that probably helped in the adoption of Microsoft servers is their SPLA agreement.  The SPLA agreement is designed for web hosters in mind.  You don’t actually have to buy the licenses for the servers you have, you simply pay a monthly fee to license the ones you need.  This allows you to expand and contract much easier with demand for servers, and your company grows.  I think this is a factor that allows businesses to use Microsoft technology for hosting much more cost effectively.

So why then did Apache drop so much in popularity?  This is a bit of a mystery to me.  Apache certainly didn’t get any less secure as a platform.  It also probably has gotten a bit easier to use, and more well documented.  Apache is also still free for anyone to install and run.  The only uncertainty for Open Source software that might cause a decline for businesses, is legal threats.  There are a few legal challenges to some of the open source technologies.  But, I don’t see this as much of a threat, so I don’t believe it would cause this much of a drop. 

The one to watch

Google is on the rise.  In fact, according to this graph Google’s service came online somehwhere during this summer.  This seems to coincide directly with Google Apps launch.  The gains are also directly related to the increase in Blogger accounts.  Blogging is taking off, anyone can blog, and anyone can gain an audience doing it.  This is appealing to many people who have something to say.  Bloggers are anywhere from someone like myself who does very technical blogging, to Grandmothers who blog about their gardens.  Google Apps website creater is very crude.  But, as with anything Google does, the first version is usually very rough, then it becomes much better and more mainstream.  Don’t be surprised if you see a really clean feature filled version of Google website creater in the near future.  When this happens, expect Google to shoot up in the market share.

The most amazing part of this, is that the industry is still rather young.  Most of the possibilities haven’t even been touched upon.  I see this industry increasing much more before it eventually levels off.  This will be a fun one to watch.

Filed under: Web, Computing