Understanding How the Internet Is Structured

In order to operate , the Internet relies on maintaining a unique set of names and numbers. The names are domain names and host names , which enable the computers connected to the Internet to be identified in a hierarchy. The numbers are Internet Protocol (IP) addresses and port numbers , which enable computers to be grouped together into interconnected sets of subnetworks, yet remain uniquely addressable by the Internet.

An Internet Service Provider (ISP) will give you the information you need to set up a connection to the Internet. You plug that information into the programs used to create that connection, such as scripts to create a Point-to-Point Protocol (PPP) connection over telephone lines. See the “Using Dial-up Connections to the Internet” section later in this chapter for descriptions of the information needed from your ISP and the procedures for configuring PPP to connect to the Internet.

The following list describes the basic Internet structure in more detail:

• IP addresses — These are the numbers that uniquely define each computer known to the Internet. Internet authorities assign pools of IP addresses (along with network masks, or netmasks ) so that network administrators can assign addresses to each individual computer that they control. An alternative to assigned addresses is to use a reserved set of private IP addresses.

  Cross-reference:

  See Chapter 15 for a description of IP addresses.

• Port numbers — Port numbers provide access points to particular services. A server computer will listen on the network for packets that are addressed to its IP address, along with one or more port numbers. For example, a Web server listens to port 80 to respond to requests for HTTP content.

• Domain names — On the Internet, computer names are organized in a hierarchy of domain names and host names. If you want to have and maintain your own Internet domain, you need to be assigned one that fits into one of the top-level domains (domains such as .com, .org, .net, .edu, .us, and so on).

• Host names — If a domain name is assigned to your organization, you are free to create your own host names within that domain. This is a way of associating a name (host name ) with an address (IP address). When you use the Internet, you use a fully qualified domain name to identify a host computer. For example, in the domain handsonhistory.com , a host computer named baskets would have a fully qualified domain name of baskets.handsonhistory.com .

  Within an organization, you should choose a host-naming scheme that makes sense to you. For example, for handsonhistory.com , you could have host names dedicated to different crafts ( baskets , decoys , weaving , and so on). Some organizations use the names from Norse mythology such as thor, odin, and loki, or beer brands such as summit, jamespage, guinness, and so on.

  Tip

  For a list of naming schemes, see ttp:// c2.com/cgi/wiki?NamesGivenToComputers .

• Routers — If you have a LAN or other type of network in your home or organization that you want to connect to the Internet, you can share an Internet connection. You do this by setting up a router. The router connects to both your network and the Internet, providing a route for data to pass between your network and the Internet. This is especially useful if you connect to the Internet over a dial-up or broadband connection, since your Linux box can act as a router and you can share the connection among all your computers.

• Firewalls and IP masquerading — To keep your private network somewhat secure, yet still allow some data to pass between it and the Internet, you can set up a firewall. The firewall restricts the kind of data packets or services that can pass through the boundary between the private and public networks. If your network uses private addresses, or if you just want to protect the addresses of computers behind your firewall, you can use techniques such as Network Address Translation (NAT) or IP masquerading.

  Note

  Although you can set up a firewall to filter packets on any computer on your private network, firewalls are typically configured most stringently on the machine that routes packets between the public and private networks. In this way, intruders can be stopped before they get on your private network and security can be relaxed somewhat between your computers behind the firewall.

• Proxies — You can bypass some of the configuration required to allow the computers on your LAN to communicate directly with the Internet by configuring a proxy server. A proxy server can store (referred to as caching ) Internet objects (such as data from Web and FTP servers) so that clients of that proxy server don't have to contact the server originating that data each time it is requested . With a proxy server, a computer on your LAN can also run Internet applications (such as a Web browser) and have them appear to the Internet as if they are actually running on the proxy server.

  Cross-Reference:

  You can read about firewalls in Chapter 14. IP masquerading and proxy servers are described in the “Enable forwarding and masquerading” and “Setting up Fedora as a Proxy Server” sections later in this chapter.

Internet domains

You can’t read a magazine, watch a TV commercial, or open a cereal box these days without coming across a “ something .com ." When a company, organization, or person wants you to connect to them on the Internet, it relies on the uniqueness of its particular domain name. However, within that domain name, the company or organization to which it has been assigned can arrange its content however it chooses.

Internet domains are organized in a structure called the domain name system (DNS). At the top of that structure is a set of top-level domains (or TLDs). Some of the top-level domains are used commonly in the United States, although they are available for worldwide use. TLDs such as edu (for colleges and universities), gov (for United States government), and mil (for United States military sites) were among the most used TLDs in the early Internet. In more recent years , com (for commercial sites) has experienced the most growth.

The us domain was added to include U.S. institutions, such as local governments and elementary schools , as well as to individuals within a geographical region of the United States. Recently, new domains such as info (for people and business to publish information about themselves ) and biz (an alternative to com for businesses) have been added.

To facilitate the entry of other countries to the Internet, the International Organization for Standardization (ISO) has defined a set of two-letter codes that are assigned to each country. Within each country are naming authorities responsible for organizing the subdomains. Some subdomains are organized by categories, while others are structured by geographic location.

Common top-level domain names

Of the generic TLDs in use today, several are used throughout the world, while two are available only in the United States. Here are descriptions of common TLDs:

• com — Businesses, corporations, and other commercial organizations fall into this TLD. As the Internet has grown into an important tool for commerce, domains in this TLD have grown at a dramatic rate.

• edu — Colleges and universities fall under this TLD. Although it was originally intended for all educational institutions, two-year colleges, high schools, and elementary schools are now organized by location under country codes (such as US in the United States).

• gov — This TLD is used primarily for U.S. federal government locations. Although most local government sites are expected to fall under the us domain, some states (including Washington state) are making the .gov TLD available to local cities and counties.

• int — This domain includes international databases and organizations created by international treaties such as NATO.

• mil — U.S. military organizations fall under this domain.

• net — Computer network providers fall under this domain.

• org — A variety of organizations that are neither governmental nor commercial in nature fall under this catchall TLD.

As noted earlier, other TLDs have been added recently to relieve some of the drain on .com names. In particular, those doing business on the Internet can get a .biz name. If you want to create a gathering point for information on a subject, you might choose a domain name from the info TLD.

Domain-name formation

As noted earlier, domain names are hierarchical, which means there can be subdomains beneath second-level domains, as well as host computers. ( Second-level domains are the names directly below the TLDs that are assigned to individual people and organizations.) Each subdomain is separated by a dot ( . ), starting with the top-level domain on the right and with the second-level domain and each subsequent subdomain appearing to the left. Here is an example of a fully qualified domain name for a host:

baskets.crafts.handsonhistory.com

In this example, the top-level domain is .com . The second-level domain name assigned to the organization that controls the domain is handsonhistory . Within that domain is a subdomain, or third-level domain, called crafts . The last name ( baskets ) refers to a particular computer within that third-level domain. From other hosts in the third-level domain, the host can be referred to simply as baskets . From the Internet, you would refer to it as baskets.crafts.handsonhistory.com .

Host names and IP addresses

In the early days of the Internet, every known host computer name and address was collected into a file called HOSTS.TXT and distributed throughout the Internet. This quickly became cumbersome because of the size of the list and the constant changes being made to it. The solution was to distribute the responsibility for resolving host names into IP addresses to many DNS servers throughout the Internet.

To make the domain names friendly, the names contain no network addresses, routes, or other information needed to deliver messages. Instead, each computer must rely on some method to translate domain names and host names into IP addresses. The DNS server is the primary means of resolving the names to addresses. If you request a service from a computer using a fully qualified domain name (including all domains and subdomains), the request will go to a DNS server to resolve that name into an IP address. It will gather that information either directly from the DNS server that owns that information or, which is more likely, from another DNS server along the way that has gathered that information.

If you have a private LAN or other network, you can keep your own list of host names and IP addresses. For the computers you work with all the time, it’s easier to type baskets than baskets.crafts.handsonhistory.com . There are a couple of ways (besides DNS) that your computer can resolve the IP address for computers for which you give only the host name:

• Check the /etc/hosts file. In your computer’s /etc/hosts file, you can place the names and IP addresses for the computers on your local network. In this way, your computer doesn’t need to query the DNS server to get the address (which may not be there anyway if you are on a private network and don’t have your own DNS server).

• Check specified domains. You can specify that if the host name requested doesn’t include a fully qualified domain name and the host name is not in your /etc/hosts file, then your computer should check certain specified domain names.

On your Fedora system, when you make a request to resolve a host name into an IP address, the contents of the /etc/resolv.conf file will most likely determine where your computer searches for that information. That file can specify your local domain, an alternative list of domains, and the location of one or more DNS servers. Here is an example of an /etc/resolv.conf file:

domain crafts.handsonhistory.com search crafts.handsonhistory.com handsonhistory.com nameserver 10.0.0.10 nameserver 10.0.0.12

In this example, the local domain is crafts.handsonhistory.com . If you try to contact a host by giving only its host name (with no domain name), your computer can check in both crafts.handsonhistory.com and handsonhistory.com domains to find the host. If you give the fully qualified domain name, it can contact the name servers (first 10.0.0.10 and then 10.0.0.12 ) to resolve the address. (You can specify up to six name servers that your computer will query in order until the address is resolved. The total search line is limited to 256 characters , however.)

If your system uses DHCP, where another server on your network assigns your Linux system an IP address, your /etc/resolv.conf file can look more like the following:

; generated by /sbin/dhclient-script search ce1.client2.big_isp.com nameserver 10.0.0.10 nameserver 10.0.0.12

In this example, the /etc/resolv.conf file was created by the DHCP client code, based on information from the DHCP server. Note that big_isp.com is an alias for a large communications company.

Routing

Knowing the IP address of the computer you want to reach is one thing; being able to reach that IP address is another. Even if you connect your computers on a LAN, to have full connectivity to the Internet there must be at least one node (that is, a computer or dedicated device) through which you can route messages that are destined for locations outside your LAN. That is the job of a router .

A router is a device that has interfaces to at least two networks and is able to route network traffic between the two networks. In my example of a small business that has a LAN that it wants to connect to the Internet, the router would have a connection and IP address on the LAN, as well as a connection and IP address to a network that provides access to the Internet.

A computer running Linux can act as a router between any two TCP/IP interfaces, for example, if the computer has two LAN cards or if it has a network interface card and a modem (for a dial-up connection to the Internet). Alternatively, you can purchase a dedicated router, such as Cisco ADSL routers, that can exclusively perform routing between your LAN and the Internet or network service provider.