Reflections on Science

by Alberto Viotto    Articles index

 

The 100 Million Names of Internet

March 2009

• First steps

• How many addresses for Internet?

• How many words to describe the world?

• The competition for the names

First steps

The beginning of the research that led to Internet dates back to 1969, when the U.S. government created the Advanced Research Projects Agency (ARPA), in order to create a network of computers that could continue to operate after a nuclear bombing. The network became active on September 2nd 1969 under the name ARPANET, connecting four universities.
In this project communication between computers was using a new technique, called packet switching. At that time, two computers required a direct connection to talk to each other. Using the new technique, messages exchanged between computers were divided into fixed-length packets and each packet was able to travel on the network by itself. In this way a packet was able to use different networks connected together and to take alternative routes to get to its final destination, even if a direct connection was not available.
The definition of protocols for packet-switched (including “tcp-ip”), dates back to those years and was formalized in 1974 in an article by Vincent Cerf and Robert Kahn. In subsequent years ARPANET became increasingly important for universities all over the world. In 1983 the Department of Defense created a separate network, MILNET, leaving ARPANET available for scientific activities. In 1986 ARPANET was replaced by a new, more capable infrastructure, called NSFNET. The first NSFNET backbone had a capacity of 56 Kbps (that is, it was able to send about 56,000 binary digits per second). For the time that was a remarkable value, but now it is the speed that a cheap PC uses to get connected to the network. In 1988, the speed of the main backbone went up to 1.5 Mbps (approximately 1,500,000 binary digits per second), in 1995 to 155 Mbps and since then has continued to increase.
The major advantage of Internet is its ability to allow any kind of device to communicate with each other, from PCs to huge computers hosted in data centers, using any type of connection, telephone cable, coaxial cable, fiber cable, satellite, radio waves, etc. The flexibility of the tcp-ip protocol is one of the reasons of the success of Internet, that now connects almost all the computers of the world (that are more than 500.000.000).

 

How many addresses for Internet?

Every computer connected to Internet is identified by its tcp-ip address, a 32 bit number. There are 4 Giga (4.294.967.295) addresses. For ease of use the tcp-ip address is split into four bytes, the value of which, expressed in decimal format, goes from 0 to 255. An example of a tcp-ip address has the form:

 

31.55.120.18

 

The tcp-ip addresses are not assigned individually, but are grouped in classes. There are three types of classes, called 'A', 'B', 'C'. In type 'A' classes all the addresses share the value of first byte. A type A class, for example, can be:

 

18.xx.xx.xx

 

This class includes 16,777,216 tcp-ip addresses, ranging from 18.0.0.0 to 18.255.255.255. Type A classes are very valued and each one was assigned to a large company or organization, such as IBM or Hewlett-Packard (which in fact has two, having inherited one from Digital Equipment). Type A classes use half of the available addresses, from 0.0.0.0 to 127.255.255.255, excluding addresses reserved for special network functions. Type 'B' classes contain addresses where the first two bytes have a fixed value, for example:

 

160.20.x.x

 

Each type B class contains 65.536 addresses. Type B classes use a quarter of the available addresses, from 128.0.0.0 to 191.255.255.255. All type A and type B classes have been assigned, while there are still type C classes available. Those classes contain addresses where the first three bytes have a specific value, such as:

 

200.24.40.x

 

Each type C class includes 256 addresses. Type C classes use the remaining addresses, from 192.0.0.0 to 255.255.255.255.

 

How many words to describe the world?

Tcp-ip addresses are difficult to remember, like phone or credit card numbers. To identify each computer on the Internet, tcp-ip addresses can be linked to logical names, which each organization can choose, with some limitations. The main requirement for this nomenclature is uniqueness. For this reason a central infrastructure was created, one of the few hierarchical components of Internet. The organization that coordinates the address assignment and translation is called ICANN (Internet Corporation for Assigned Names and Numbers). ICANN manages the so-called "top-level domains" and delegates the management of lower-level domains to other organizations.

Top-level domains represent a first subdivision of Internet computers, according to the type of organization they belong to, whether commercial, academic, military. The main top-level domains are:

 

 

As an alternative to this subdivision, you can use a country identifier. Every country in the world has been assigned a unique code of two letters, which usually coincides with the first two letters of its name in its official language, for example:

 

 

There are  240 national codes. Using two letters, each of which can take 26 values, you could get 676 possible combinations. In the event that two nations should use the same code (for example, "Austria" and "Australia"), the second letter of the code must be changed:

 

 

The central authority is not directly managing the various domains, but delegates other organizations. The domain "it", assigned to Italy, is managed by the CNR, located in Pisa. Within each top-level domain, many “second-level domains” can be defined. Each second-level domain defines a specific business or organization. The company "Frescopresto", for example, could request the authority which manages “.com” this domain:

 

frescopresto.com

 

or could request the Italian authority this domain:

 

frescopresto.it

 

These two domains may also be assigned to two different companies. Problems of homonymy may arise, especially for the domain "com", the most crowded, which contains about 50 million second-level domains (while the total number of registered domain is close to 100 million). The majority of short names, which are easy to remember, have already been assigned and newcomers will only be able to get long names.

 

The competition for the names

The competition for Internet names can be very harsh. For instance, millions of pizza restaurants would like get the domain:

 

pizza.com

 

There are speculators who keep a large number of domains (the contribution for registration is quite low) for sale to others. Bill Gates himself wrote an article joking about the fact that the domain "billgates.com" was sold for one million dollars, although usually prices are much lower. If the name you want has already been registered and you don’t want to pay its holder, you can almost always find a suitable name, adding or changing some letters. For example, if the municipality of Rome wants the domain "roma.it" and discovers that it has already been assigned, they can choose from a wide range of alternatives:

 

cittadiroma.it
romacapitale.it
comunediroma.it
comune-roma.it
romaurbe.it
...

 

The overall index of second-level domains contains all the entities that need to be identified on the Internet, all the companies and organizations in the world. To describe the world two words are enough, one to identify the top-level domain and another for the second-level domain.

 

     Alberto Viotto

 

The Italian page (including italian articles, a free ebook about paradoxes, a video of an interview about the ebook “Computers’ revenge”): Riflessioni sulle Scienze