A week can be a long time on the internet, so fast is the pace of innovation; but as with a lot of other reporting, it often takes a war for the media to take notice. And what a week it's been...
The story begins with the Wikileaks organisation, publishers of leaked documents from whistleblowers everywhere, coming to international attention once again with the publication of leaked diplomatic cables from the US Government. Then on 2nd December, the Wikileaks website - wikileaks.org - seemed to disappear from the web. But where did it go?
At the time, it didn't actually go anywhere... it was just hidden to most of us. The web addresses we're all familiar with, www.open.ac.uk for example, or wikileaks.org don't actually tell your browser where a particular website lives. Instead, the 'domain names' are aliases - memorable and human readable labels used to mask the physical address of a particular web server (or family of webservers) somewhere on the internet.
These machine readable IP addresses take the form of rather less memorable sets of numbers arranged in a particular and well defined way: 126.96.36.199 is the address of the Open University website, for example (try it: http://188.8.131.52).
The same goes for Wikileaks, which could be found at http://184.108.40.206 when it seemed to disappear (although a whole host of other websites are mirroring the site; which is to say, there are now copies of the Wikileaks site all over the web!)
When a browser wants to fetch a web page, it must first of all identify the numeric IP address of the webserver that the page is hosted on from its domain. This is done by looking up the name using the domain name service - DNS. The OpenLearn Unit "Protocols in Multiservice Networks" illustrates the process as follows:
A lookup, such as request for the numerical IP address equivalent of 'www.open.ac.uk' is referred to a root level name server, which passes the request on to a 'uk' level name server. Requests are passed to ever more specific name servers until at last one is found that knows the actual address. You might think of this a bit like the way the Post Office handles international mail; individual items of post are passed to ever more specific sorting offices, terminating with the postman who actually has enough local knowledge to be able to deliver a particular letter to a specific physical address.
What happened when Wikileaks seemed to disappear was that the provider of its domain name service removed its address information from their servers. This meant that when a request to look up the actual physical address was made, there was no 'wikileaks.org' nameserver that could divulge the physical internet address of the Wikileaks server. As a result, browsers couldn't connect to the actual Wikileaks webserver because they couldn't find out where to connect to it just given the 'wikileaks.org' domain name. But the Wikileaks server was still there, and still reachable if you knew the numerical IP address of the server.
In fact, within hours, the 'dotted quad' numerical address for the Wikileaks servers was being widely circulated, and to all intents and purposes, Wikileaks became discoverable again.
But why does the internet rely on this two tiered approach to addressing? One reason is that it lets you move a website from one physical server to another. At a certain point during last week, Wikileaks content was being served from Amazon web servers (you may be surprised to learn that Amazon.com is not only a online retailer, it also provides a whole range of infrastructure level web services for some of the biggest web properties in the world). At least, until Amazon determined that Wikileaks was in breach of its Terms and Conditions, and threw Wikileaks off.
This meant that Wikileaks had to change physical servers to ones with new physical addresses. In the normal course of events, this would have just meant changing the address information in the domain name service, because the human readable domain name (wikileaks.org, for example) masks the numerical address. So we can change the numerical address and still use the same domain name.