History Overview of DNS

DNS can stand for Domain Name Service (DNS for short), or Domain Name Server (nameserver for short). DNS is the distributed Internet service that provides the resolution from name based Internet addresses (domain names and host names), to a domain name or host (a machine, aka node)'s actual IP addresses.

TCP/IP is usually what is used when describing the Internet protocol's, but it's really just IP. Although most Internet traffic uses TCP transport protocols over the IP network protocol, there are other transport protocols as well (such as UDP, ESP, PPTP).

Let's first take a look at some more traditional network protocols.

The United States Postal Service

If you want to send a letter to someone, you must place the destination address on the letter in order for the postal service to know how to route the letter to it's intended recipient. Most mail doesn't pass directly between two locations. A piece of mail is delivered to your local postal service who decides to which of the 1000's of postal substations to route it to in order to get the letter going in the right direction towards its destination. They use the ZIP code to help them to determine this. Once they deliver the mail to the next substation, on its way to its final destination, their job is over. They assume that the next substation will pass it on to the next substation, and so on and so on, until the letter gets to where it's going.

Don't forget, you also need to put a return address on the letter so that its recipient will have an address to which to reply.

IP Network Protocol

Sending data over the Internet works in much the same way as the US Postal Service does. If you want to send data to a remote location, you place it's destination address on the data and deliver it to your local Internet router.

This first Internet router determines, by the IP address, to which other router it should send to get the data going in the right direction towards its intended recipient. Once the router delivers the data to the next router, its job is done. It expects for each successive router that the data is delivered to will responsibly get it to its final destination by continuing to route it along.

IP addresses are made up of numbers for the same reason that zip codes are. It makes it easier, using simple hash maps, to determine the route between two addresses. Also, digital computing devices (such as the popular PC) typically use binary because its transistors manipulate numbers as strings of ones and zeroes (which stand for on and off). IP addresses under IPV4 (the current IP standard) are 4 bytes long. Each byte is 8 bits, so an IP address consists of 32 bits. The first 24 bits usually represent the "zip code", which the Internet protocol calls the network address. Typically, the last 8 bits are used as the "street address", which the Internet protocol calls the node address (aka host address). So take a look at the following IP address:

11000000101010000000000000000001  <--- 32 bit IP address
110000001010100000000000  <--- 24 bit Network Address
                        00000001  <--- 8 bit Node Address

This IP address can be shown in a more human readable form known as the dotted decimal notation format. This format splits the address up into its four bytes, shows each octet in its base 10 decimal equivalent, and uses a decimal point (better known as a dot) to seperate the four bytes.

11000000101010000000000000000001  <--- 32 bit IP address
192.168.0.1  <--- In Dotted Decimal Notation
192.168.0  <--- Significant network bytes (24 bits)
192.168.0.0  <-- Dotted Decimal Network Address
          1  <-- Significant node byte (8 bits)
192.168.0.1  <-- Dotted Decimal Node Address

There is another part to this puzzle known as the netmask, that determines how many bits of the 32 bit address should be considered as the network address, the remainder representing the local node address. We won't cover this a lot, but suffice it to say that we usually assume that people are using a 24 bit netmask, which is a "C" class network in class based IP networking.

So in order for you to send data to a host on the Internet, to receive data back as a reply, or to allow data(visitors) to find their way to your host, people need to use this 32 bit IP address. To make things even easier to remember, Al Gore came up with the Domain Name System, where a network could be represented with a domain name, and a node with a host name. So you could name your network "mydomain", and give your document server the name "www" (you might name it this because you'll be using to serve documents over the Internet usin HTTP and the World Wide Web, or just because everyone else does it). The host name is indicated before the domain name, with a dot separating the two. So your world wide web server's full name based address would be "www.mydomain". In order for a message to be deliverable over IP, this name based address still needs to be converted into an IP address. This is where Domain Name Service comes in. DNS provides a service where you can query a server for the IP address that a name should resolve to.

The Internet's dozen or so root servers provide the resolution of domain names to IP addresses. For logistic reasons, however, it was determined that these servers would not be required to perform resolution for every single hostname on every single network. Rather, it is necessary that each network get their own local DNS servers for the resolution of the hostnames on their network. The root servers' only job is to answer queries with the IP addresses of the nameservers that maintain the authoritative record for any particular domain name (network). You can then use those IP addresses to query that network's local DNS servers to resolve host names under it's domain.

Note: The root servers, and those who maintain them, are often referred to as the InterNIC, for reasons that are no longer so obvious.

Because we need to control the assignment/use of IP addresses and domain names, just as the US Postal Service assigns Postal Addresses to it's customers, several controlling organizations were formed to do the same for Internet addresses. To better distribute the assignment of domain names, managers were created for the registry of Top Level Domains. Two-letter TLD's were created for country codes, as well as the common three-letter TLD's for organizations and business entities. Network Solutions was the first company to win a contract for managing the assignment of the .net, .com, and .org TLD's. If you wanted to register a domain name in one of these TLD's you could do it through Network Solutions. Which ever TLD your address was placed in would be appended to your name based address. So, "www.mydomain" would become "www.mydomain.com".

Today, the control of the .net, .com, and .org TLD's has been transferred to the Internet Corporation for Assigned Names and Numbers (ICANN). Many companies have applied and have been approved as ICANN Accredited Registrars. These organizations can handle registering domain names in the .net, .com, and .org TLD for you through there affiliation with ICANN.

The Companies Involved

The American Registry for Internet Numbers (ARIN) controls the assignment of IP addresses for the Americas.

The Internet Corporation for Assigned Names and Numbers (ICANN) is responsible for the assignment of domain names under TLD's. Domain names under the .com, .net, and .org TLD's are registered through one of their approved registrars. Domain names under two-letter contry code TLD's are done through the managing organization for that country's TLD, which can also be found through ICANN's web site. These managing organizations must also be approved by ICANN.

The Local Network Administrator will normally be in charge of the authoritative name servers for a domain name. This person is responsible for ensuring that there are at least two nameservers permanently connected to the Internet to provide the resolution of host names under it's domain to IP addresses. The nameserver does this by having records for each domain (aka zone) in a zone file. The nameserver reads these zone files upon startup and caches the entries so that it can respond to queries faster. The primary nameserver is usually the master, and the secondary nameserver is a slave. This means that they both have the same zone files, but you only have to update the zone records on the master. Periodically, the slave will ask the master if there have been any changes by comparing the serial number at the beginning of the zone file to see if it matches its own copy of the zone file. If they don't match, the slave will do a zone transfer to update its zone file for that zone.