 The magazine of the Melbourne PC User Group A hitchhiker's guide to domain names Peter Stanski and Deshrani Rodrigo |
Six years ago, about one million host computers and just 6500 networks
were part of the Internet. Now there are over 20 million host computers and more than 1.3 million networks.
There are also approximately 60 million Internet users worldwide, and this number continues to more than
double each year.
All these hosts are currently operational, sending and receiving billions of packets each and every second.
Driven by the Transmission Control Protocol/Internet Protocol (TCP/IP), each machine communicates with its
neighbours by an IP address. These are often written down in a dotted decimal notation. Something like
130.194.224.20, is the unique IP address for an Internet host.
The Internet Assigned Numbers Authority (IANA), has been responsible for allocating and maintaining unique IP
addresses to new machines appearing on the Net. Despite the fact that an IP address is a 32-bit number, an IP
address shortage is occurring. The currently used version of IP, is version 4 (IPv4--four decimal notations),
and a new proposed version is around the corner to overcome the addresses deficit. Originally known as IP
Next Generation, IP version 6 (IPv6--16 decimal notations for 128-bit address) will provide a larger number
of unique IP addresses and also extend the protocol's functionality.
Using names as addresses
Most of us have significant problems remembering people's phone numbers and therefore are unlikely to recall
an IP address with ease. Consequently, the Internet Engineering Task Force (IETF) with some help from the
people at Berkeley, introduced the Domain Name concept. Their idea was to provide a naming system which could
easily resolve a host name, to its IP address. This means that some hosts can be given one or more names,
which are aliases for its IP address. Similarly, one host name may map to several IP addresses if a need for
this is anticipated.
When a user wishes to communicate with a remote machine, all that is required is the host's name. These are
known as Internet Domain Names, which are unique across the Internet. This means that any given Internet
Domain Name such as bambi.com.au and aqua.ct.monash.edu.au are unique names for all of the
Internet.
One central organization has been associated with the job of registering and maintaining every domain name.
In the US, it is known as InterNIC (Network Information Center) which consists of three contracted companies,
General Atomics (GA), AT&T and Network Solutions Inc. (NSI). InterNIC is involved with other
international domain name maintainers for each respective country. These are often the country's largest
Internet Service Providers (ISPs). Each has to collaborate with the other to a small degree, by delegating
names to special domains. For instance, MelbourneIT handles domain name allocations in the .au
domain.
That is great so far, but how does your computer know of all the other millions of names that are in use, or
the one that you are after?
Domain name servers
This problem was solved by having a central Domain Name Server (DNS). For historical reasons, and the lack of
TCP/IP in many early operating systems, a UNIX system was often chosen to hold a large database of aliases
and IP addresses.
When a host needs to resolve a hostname to an IP address, it simply asks a local DNS host. It knows whom to
ask, since the IP address of a DNS node is globally known to all nodes. Without knowing the IP of the DNS, it
would be almost impossible to resolve all Internet hostnames to their IPs.
Once contacted, the DNS server accepts a hostname and returns its IP address(es) to the node which requested
the name to address resolution. Once an IP address has been found, the originating node may establish a
connection to the remote machine.
This concept has proven to be a total success. Because of its simplicity and transparency to the user, it is
used by every Internet application. However, some casual Internet users are often unaware that the Internet's
TCP/IP protocol operates using IP addresses alone, and not by the remote host's name. As users work with
names, some may be oblivious to the fact that these names are resolved, giving IP addresses with which the
TCP/IP protocol operates. As mentioned previously, names are only utilised to make things easier for
users.
Reverse DNS
Having a way to convert a name to an IP address is fundamental, but converting an IP address to a name
involves security issues that need to be addressed. With the Internet being a large network, it is most
likely that there are some who will attempt to abuse the system. These so called hackers, are likely to know
many tricks to penetrate an insecure system.
Hence, to assist in Internet security, a Reverse Domain needs to be created. This is the reverse of what a
DNS server performs. A host which receives a connection from a remote host with a certain IP address, may
look up the caller's domain name and then decide whether to accept the connection. This is similar to
screening incoming phone calls by seeing where the caller appears to be at. This constitutes a simple level
of security by allowing only connections from trusted domains to communicate.
Locating hosts by names
A name given to a host, is called a hostname. It is written using a dot notation similar to IP decimal
notation, which represents the hostname and its network domain. A sample hostname called Bambi, which is a
machine used for commercial purposes appears as bambi.com.au.
When reading domain names, these should be read from right to left. The final part of the name, being ".au",
represents that the machine's physical location is in Australia. The second last part, "com", represents that
the machine is used for commercial purposes. Finally, the left most "bambi", is the name given to that
particular node. Some hostnames may be longer than this example, but the same decoding rules apply when
reading multi-level domain names.
A DNS server reads these domain names and decides which database holds the international entries. As it is
impossible for one DNS server to hold all domain names, it calls upon its parent server to resolve some
names. Such name resolution delegation may occur several times until a DNS server is reached internationally,
which holds the correct hostname entry.
Internet users may have noticed that the first time a connection to a host is established it may take
several seconds, while subsequent connections are faster. This is because the first connection requires the
DNS hostname look-up internationally, while subsequent ones are looked up from a local cache which has been
created for future access.
It should also be noted that all domain names have the right most field as their International Standards
Organization (ISO) two-letter standard country name. This applies to all countries except for the Unites
States, who do not have this. Therefore, domain names without a specified country, are generally identified
as being within the US.
Domains by function
We can also establish other things from a domain name. Earlier we mentioned that bambi.com.au was used
for commercial purposes and that it was located in Australia. As noted previously, the domain name after the
two letter country code (reading right to left), is listed as an Australian commercial domain by
".com.au".
All registered domain names are required to be placed in one of six organisation domains; NET for network
operation centers, MIL for military, GOV for government, COM for commercial, EDU for educational, and ORG for
organizations which do not fall into the other five categories. These allow users to establish more
information about a site by merely examining its domain name.
New domain names
Along with IP address shortages, there have also begun to appear some domain name shortages. Just as names of
individuals in the world are not unique, there are also many organizations which have the same name and
belong to the same domain. Although the possible number of domain names is virtually infinite, as it can
contain any unique combination of three to sixtythree characters, organisations with the same names compete
for meaningful domain names as these are limited (www.gtld-mou.org).
The Internet Ad Hoc Committee (IAHC) has suggested new domains be added to the existing six. The
introduction of these is likely to be discussed and decided upon in the not too distant future. The addition
of new top-level domains will be an effective way of solving the current problem of a shortage of domain
names.
This will also ease the current competition between organizations for the most popular domain name. This
competition is comparable to obtaining the coolest customized license plate, or a memorable phone number for
a new cellular phone. In addition, domain names also have economic and intellectual property values. This is
due to extensive advertising that associates and identifies many companies with their respective domain
names. It is much more meaningful for companies to advertise using domain names, such as
www.microsoft.com instead of the IP address 203.17.138.84, which customers are unlikely to remember, let
alone associate with the company.
Multi-level domains
So far we have described the top-level name domains. These are used as the first criteria for establishing a
host's role. For example, educational institutions would be placed in the .EDU domain, but after that
classification is complete, it is up to the network administrators of a site to create their own sub-domains.
Subdomains are often found in large scale organisations/institutions wishing to manage their hosts on a
departmental, business unit, or geographic basis.
A sample complete Internet domain name for a machine looks something like: aqua.ct.monash.edu.au.
Reading from right to left, the machine is located in Australia, falls into the educational domain (academic
in this case), is owned by Monash University, is within the department of Computer Technology (CT), and
finally the hostname is Aqua.
More experienced users can establish, and often even guess names when only a partial name is known. This is
similar to guessing company sites by prefixing WWW, dot, company name, dot, COM (WWW.company.COM).
This aids those who have troubles remembering machine names. Knowing the structure of domain names can
sometime help in locating a site you have once visited, but can not quite remember.
Choosing a hostname
Users are seldom involved in choosing a hostname for their Internet machine. These are often provided by your
Internet Service Provider (ISP) who may be someone you call from home over your modem, or the network
administrator in your organisation.
Hostnames greatly aid in communicating a problem to your provider. For example, let us say that your
workstation called bambi.ct has broken down. All you have to say to your technicians is that
bambi.ct is down and needs to be fixed. This is far more meaningful than saying that your computer has
broken down and you are on the 3rd floor in E-Block.
Some organisations may even use more meaningful hostnames, identifying the location of machines. These names
may use building numbers/names, floor numbers and even room numbers. An example of such a hostname is
HP5L-C320.ct. This relates to a HP5L printer in C-Block, level 3, room 20, owned by the CT
department.
Those fortunate enough to be given the privilege of choosing their own name for a machine may wish to look
in RFC1178--"Choosing a Name for Your Computer", for some simple guidelines.
Web names
The World Wide Web (WWW) has affected a great deal of domain names. Most systems in the early days of the Web
offered the web service just like any other Internet service. However, establishing which systems did or did
not offer web services was difficult. To discover the availability of a service a user first had to attempt
to connect to it.
This led to a naming scheme, where machines which were offering the WWW service were given "WWW" as their
host name prefix. Hence, domain names like www.intel.com or www.microsoft.com became increasingly
popular, the hostname defining the exclusive service of the machine for the given organisation.
The decision to name the hostname after its exclusive service has been around since DNS appeared. Internet
mail hubs, news-servers, gateways and routers have also had their names used to explicitly describe their
services.
Conclusion
The continual growth of the Internet has brought about many challenges and opportunities. In this article we
focused on the issues pertaining to addressing, concentrating mainly on domain names. The logic behind the
concept of using names, as opposed to IP addresses was considered from a user's perspective.
Accordingly, the need for name resolution using DNS Servers, as well as the benefits of reverse DNS were
briefly discussed. The latter discussion was motivated by security issues that we do not often think about,
yet are important to maintaining the integrity of our systems.
Domain names were also dissected in this article, taking a closer look at their parts, and discovering that
they are more meaningful than what they seem to be at first glance. It was found that not only can they be
read (from right to left), but that the names also give a lot of information about their functions and usage.
Large companies and institutions exploit this feature of domain names, using multi-level domains to better
organise their workplace, by giving easy to remember, meaningful hostnames.
Inevitably, like the IP address shortage, the growth and popularity of domain names has lead to a shortage
of these too. Currently, this problem is being discussed by various forums, with a solution expected in the
not too distant future.
About the authors
Dr Peter Stanski...oops I mean Peter Stanski (darn, still have not finished that PhD yet!) is an
over-qualified computer nerd whose idea of a great overseas trip is to visit bookstores in Silicon Valley,
just so that he can purchase all the latest computer books. He can be reached at: peter.stanski@csse.monash.edu.au
Deshani Rodrigo is a movie going funster, turned computer geek recently. In her spare time she watches every
sitcom under the sun, shops and occasionally reads boring computer books so she can finish her Masters
degree. Her contact is
drodrigo@insect.sd.monash.edu.au
Reprinted from the June 1999 issue of PC Update, the
magazine of Melbourne PC User Group, Australia
