DNS scales well. It doesn't rely on a single large table; it is a distributed database system that doesn't bog down as the database grows. DNS currently provides information on approximately 16,000,000 hosts, while less than 10,000 are listed in the host table.
DNS guarantees that new host information will be disseminated to the rest of the network as it is needed.
Information is automatically disseminated, and only to those who are interested. Here's how it works. If a DNS server receives a request for information about a host for which it has no information, it passes on the request to an authoritative server. An authoritative server is any server responsible for maintaining accurate information about the domain being queried. When the authoritative server answers, the local server saves (caches) the answer for future use. The next time the local server receives a request for this information, it answers the request itself. The ability to control host information from an authoritative source and to automatically disseminate accurate information makes DNS superior to the host table, even for networks not connected to the Internet.
In addition to superseding the host table, DNS also replaces an earlier form of name service. Unfortunately, both the old and new services are commonly called name service. Both are listed in the /etc/services file. In that file, the old software is assigned UDP port 42 and is called nameserver or name. DNS name service is assigned port 53 and is called domain. Naturally, there is some confusion between the two name servers. This text discusses DNS only; when we refer to "name service," we always mean DNS.
DNS is a distributed hierarchical system for resolving hostnames into IP addresses. Under DNS, there is no central database with all of the Internet host information. The information is distributed among thousands of name servers organized into a hierarchy similar to the hierarchy of the UNIX filesystem. DNS has a root domain at the top of the domain hierarchy that is served by a group of name servers called the root servers.
Just as directories in the UNIX filesystem are found by following a path from the root directory, through subordinate directories, to the target directory, information about a domain is found by tracing pointers from the root domain, through subordinate domains, to the target domain.
Directly under the root domain are the top-level domains. There are two basic types of top-level domains - geographic and organizational. Geographic domains have been set aside for each country in the world, and are identified by a two-letter code. For example, the United Kingdom is domain UK, Japan is JP, and the United States is US. When US is used as the top-level domain, the second-level domain is usually a state's two-letter postal abbreviation (e.g., WY for Wyoming). US geographic domains are usually used by state governments and K-12 schools and are not widely used for other hosts within the United States.
Within the United States, the most popular top-level domains are organizational - that is, membership in a domain is based on the type of organization (commercial, military, etc.) to which the system belongs.  The top-level domains used in the United States are:
 There is no relationship between the organizational and geographic domains in the U.S. Each system belongs to either an organizational domain or a geographical domain, not both.
Several proposals have been made to increase the number of top-level domains. The proposed domains are called generic top level domains or gTLDs. The proposals call for the creation of additional top-level domains and for the creation of new registrars to manage the domains. All of the current domains are handled by a single registrar - the InterNIC. One motivation for these efforts is the huge size of the .com domain. It is so large some people feel it will be difficult to maintain an efficient .com database. But the largest motivation for creating new gTLDs is money. Now that it charges fifty dollars a year for domain registration, some people see the InterNIC as a profitable monopoly. They have asked for the opportunity to create their own domain registration "businesses." A quick way to respond to that request is to create more official top-level domains and more registrars. The best known gTLDs proposal is the one from the International Ad Hoc Committee (IAHC). The IAHC proposes the following new generic top-level domains:
businesses or firms
businesses selling goods
organizations emphasizing the World Wide Web
cultural and entertainment organizations
recreational and entertainment organizations
sites providing information services
individuals or organizations that want to define a personal nomenclature
Will the IAHC proposal be adopted? Will it be modified? Will another proposal win out? I don't know. There are several other proposals, and as you would expect when money is involved, plenty of controversy. At this writing the only official organizational domain names are: com, edu, gov, mil, net, int, and org.
Figure 3.1 illustrates the domain hierarchy by using the organizational top-level domains. At the top is the root. Directly below the root domain are the top-level domains. The root servers only have complete information about the top-level domains. No servers, not even the root servers, have complete information about all domains, but the root servers have pointers to the servers for the second-level domains.  So while the root servers may not know the answer to a query, they know who to ask.
 Figure 3.2 shows two second-level domains: nih under gov and nuts under com.
The Network Information Center has the authority to allocate domains. To obtain a domain, you apply to the NIC for authority to create a domain under one of the top-level domains. Once the authority to create a domain is granted, you can create additional domains, called subdomains, under your domain. Let's look at how this works at our imaginary nut packing company.
Our company is a commercial profit-making (we hope) enterprise. It clearly falls into the com domain. We apply to the NIC for authority to create a domain named nuts within the com domain. The request for the new domain contains the hostnames and addresses of at least two servers that will provide name service for the new domain. (Chapter 4, Getting Started discusses the domain name application.) When the NIC approves the request, it adds pointers in the com domain to the new domain's name servers. Now when queries are received by the root servers for the nuts.com domain, the queries are referred to the new name servers.
The NIC's approval grants us complete authority over our new domain. Any registered domain has authority to divide its domain into subdomains. Our imaginary company can create separate domains for the sales organization (sales.nuts.com) and for the packing plant (plant.nuts.com) without consulting the NIC. The decision to add subdomains is completely up to the local domain administrator.
Name assignment is, in some ways, similar to address assignment. The NIC assigns a network address to an organization, and the organization assigns subnet addresses and host addresses within the range of that network address. Similarly, the NIC assigns a domain to an organization, and the organization assigns subdomains and hostnames within that domain. The NIC is the central authority that delegates authority and distributes control over names and addresses to individual organizations. Once that authority has been delegated, the individual organization is responsible for managing the names and addresses it has been assigned.
The parallel between subnet and subdomain assignment can cause confusion. Subnets and subdomains are not linked. A subdomain may contain information about hosts from several different networks. Creating a new subnet does not require creating a new subdomain, and creating a new subdomain does not require creating a new subnet.
A new subdomain becomes accessible when pointers to the servers for the new domain are placed in the domain above it (see Figure 3.1 Remote servers cannot locate the nuts.com domain until a pointer to its server is placed in the com domain. Likewise, the subdomains sales and plant cannot be accessed until pointers to them are placed in nuts.com. The DNS database record that points to the name servers for a domain is the NS (name server) record. This record contains the name of the domain and the name of the host that is a server for that domain. Chapter 8, Configuring DNS Name Service , discusses the actual DNS database. For now, let's just think of these records as pointers.
Figure 3.2 illustrates how the NS records are used as pointers. A local server has a request to resolve salt.plant.nuts.com into an IP address. The server has no information on nuts.com in its cache, so it queries a root server (terp.umd.edu in our example) for the address. The root server replies with an NS record that points to almond.nuts.com as the source of information on nuts.com. The local server queries almond, which points it to pack.plant.nuts.com as the server for plant.nuts.com. The local server then queries pack.plant.nuts.com, and finally receives the desired IP address. The local server caches the A (address) record and each of the NS records. The next time it has a query for salt.plant.nuts.com, it will answer the query itself. And the next time the server has a query for other information in the nuts.com domain, it will go directly to almond without involving a root server.
Figure 3.2 is an example of a non-recursive query. In a non-recursive query, the remote server tells the local server who to ask next. The local server must follow the pointers itself. In a recursive search, the remote server follows the pointers and returns the final answer to the local server. The root servers generally perform only non-recursive searches.
Domain names reflect the domain hierarchy. Domain names are written from most specific (a hostname) to least specific (a top-level domain), with each part of the domain name separated by a dot.  A fully qualified domain name (FQDN) starts with a specific host and ends with a top-level domain. peanut.nuts.com is the FQDN of workstation peanut, in the nuts domain, of the com domain.
 The root domain is identified by a single dot; i.e., the root name is a null name written simply as "
Domain names are not always written as fully qualified domain names. Domain names can be written relative to a default domain in the same way that UNIX pathnames are written relative to the current (default) working directory. DNS adds the default domain to the user input when constructing the query for the name server. For example, if the default domain is nuts.com, a user can omit the nuts.com extension for any hostnames in that domain. almond.nuts.com could be addressed simply as almond. DNS adds the default domain nuts.com.
This feature is implemented in different ways on different systems, but there are two predominant techniques. On some systems the extension is added to every hostname request unless it ends with a dot, i.e., is qualified out to the root. For example, assume that there is a host named salt in the subdomain plant of the nuts.com domain. salt.plant does not end with a dot, so nuts.com is added to it giving the domain name salt.plant.nuts.com. On most systems, the extension is added only if there is no dot embedded in the requested hostname. On this type of system, salt.plant would not be extended and would therefore not be resolved by the name server because plant is not a valid top-level domain. But almond, which contains no embedded dot, would be extended with nuts.com, giving the valid domain name almond.nuts.com.
How the default domain is used and how queries are constructed varies depending on software implementation. It can even vary by release level. For this reason, you should exercise caution when embedding a hostname in a program. Only a fully qualified domain name or an IP address is immune from changes in the name server software.
DNS name service software is conceptually divided into two components - a resolver and a name server. The resolver is the software that forms the query; it asks the questions. The name server is the process that responds to the query; it answers the questions.
The resolver does not exist as a distinct process running on the computer. Rather, the resolver is a library of software routines (called the "resolver code") that is linked into any program that needs to look up addresses. This library knows how to ask the name server for host information.
Under BIND, all computers use resolver code, but not all computers run the name server process. A computer that does not run a local name server process and relies on other systems for all name service answers is called a resolver-only system. Resolver-only configurations are common on single user systems. Larger UNIX systems run a local name server process.
The BIND name server runs as a distinct process called named (pronounced "name" "d"). Name servers are classified differently depending on how they are configured. The three main categories of name servers are:
The primary server is the server from which all data about a domain is derived. The primary server loads the domain's information directly from a disk file created by the domain administrator. Primary servers are authoritative, meaning they have complete information about their domain and their responses are always accurate. There should be only one primary server for a domain.
Secondary servers transfer the entire domain database from the primary server. A particular domain's database file is called a zone file; copying this file to a secondary server is called a zone file transfer. A secondary server assures that it has current information about a domain by periodically transferring the domain's zone file. Secondary servers are also authoritative for their domain.
Caching-only servers get the answers to all name service queries from other name servers. Once a caching server has received an answer to a query, it caches the information and will use it in the future to answer queries itself. Most name servers cache answers and use them in this way. What makes the caching-only server unique is that this is the only technique it uses to build its domain database. Caching servers are non-authoritative, meaning that their information is second-hand and incomplete, though usually accurate.
The relationship between the different types of servers is an advantage that DNS has over the host table for most networks, even very small networks. Under DNS, there should be only one primary name server for each domain. DNS data is entered into the primary server's database by the domain administrator. Therefore, the administrator has central control of the hostname information. An automatically distributed, centrally controlled database is an advantage for a network of any size. When you add a new system to the network, you don't need to modify the /etc/hosts files on every node in the network; you modify only the DNS database on the primary server. The information is automatically disseminated to the other servers by full zone transfers or by caching single answers.
The Network Information Service (NIS)  is an administrative database system developed by Sun Microsystems. It provides central control and automatic dissemination of important administrative files. NIS can be used in conjunction with DNS, or as an alternative to it.
 NIS was formerly called the "Yellow Pages," or yp. Although the name has changed, the abbreviation yp is still used.
NIS and DNS have similarities and differences. Like DNS, the Network Information Service overcomes the problem of accurately distributing the host table, but unlike DNS, it provides service only for local area networks. NIS is not intended as a service for the Internet as a whole. Another difference is that NIS provides access to a wider range of information than DNS - much more than name-to-address conversions. It converts several standard UNIX files into databases that can be queried over the network. These databases are called NIS maps.
NIS converts files such as /etc/hosts and /etc/networks into maps. The maps can be stored on a central server where they can be centrally maintained while still being fully accessible to the NIS clients. Because the maps can be both centrally maintained and automatically disseminated to users, NIS overcomes a major weakness of the host table. But NIS is not an alternative to DNS for Internet hosts, because the host table, and therefore NIS, contains only a fraction of the information available to DNS. For this reason DNS and NIS are usually used together.
This section has introduced the concept of hostnames and provided an overview of the various techniques used to translate hostnames into IP addresses. This is by no means the complete story. Assigning host names and managing name service are important tasks for the network administrator. These topics are revisited several times in this book and discussed in extensive detail in Chapter 8.