To simplify this example, we will assume that the client node knows the complete IP address of the destination Web server. When the client sends a message to a server in another network, the client's IP process recognizes that the network portion of the packet's destination IP address does not match the network portion of the sending node's IP address. That means that the destination is on some distant network. The client's IP process does not know how to reach that distant network, thus it addresses the frame to the router that serves as its default gateway.

Thus this time, the client transmits a frame addressed to the router. However, that frame contains a packet addressed to the distant Web server. Once again, the hub broadcasts the frame, which the other clients ignore. The switch recognizes the frame address of the router, and forwards the frame only to the port connected to the router. The destination address of the frame matches the NIC in the router, thus the router's NIC sends the IP packet up to its IP process.

The router's IP process recognizes that the network portion of the destination IP address does not match the network portion of its own IP address. The router checks its routing table, and finds an entry for the destination network. That table entry tells the router what router port is connected to a link that leads to the destination network. That link may be either directly or indirectly connected to the destination; however, that port will eventually get the packet where it needs to go, and that is all the router cares about.

In this case, the correct router port is connected to a wide area link, such as a frame relay virtual circuit. The router now builds a frame relay frame to encapsulate the IP packet, then transmits the new frame over the physical wide area link. The frame relay frame is addressed to the frame relay address of the router connected to the destination network. (You will learn the details of frame relay addressing in later courses; however, that is close enough for now.)

The router at the other end of the frame relay connection removes the frame relay header and trailer, then evaluates the IP packet address. It recognizes that its network number matches the network portion of the destination IP address, thus it knows that the packet must go somewhere within its own network. To build a frame that will get the packet to the server, the router now needs the NIC address of the server.

If the router does not already know the NIC address, it broadcasts an Address Resolution Protocol (ARP) request frame. The ARP request asks for the NIC address that corresponds to the IP address. When the server broadcasts its NIC address in an ARP reply, all nodes, including the router, record that address in their ARP cache, which is a list that maps IP addresses to NIC addresses. Each node checks its cache first when it needs the NIC address of a destination.

The router then encapsulates the packet within an Ethernet frame addressed to the Web server's NIC. The router transmits the new Ethernet frame over the router port that leads to its local network. The switch forwards the frame to the switch port that is connected to the Web server. The Web server's NIC processes the frame, and the client's original Web request message is finally passed up the server's protocol stack to the Web server application.

DNS

Earlier, we simplified this example by assuming the Web client knew the full IP address of the destination Web server. In many cases this is true, especially in a corporate intranet. However, in the Web, it is rare for a client node to know a server's address. New servers are constantly added to the worldwide system, and Web content is often moved from one server to another.

The Domain Name System (DNS) provides a way to navigate the Web without knowing each server's IP address. Clients use easy-to-remember domain names, such as "westnetinc.com." A computer called a "DNS server," or name server maps these alphanumeric Internet addresses to binary IP addresses.

DNS servers (or software that provides DNS functionality) are found in virtually every network that uses TCP/IP as the networking protocol. The Internet's DNS servers regularly exchange new and changed addresses. Within a few days, a change in an Internet address is received by every DNS server around the world.

Thus, when a router receives a packet addressed to "westnetinc.com," it asks the nearest DNS server to supply the IP address that corresponds to that domain name. The name server looks up the name in its database, then returns an IP address in binary format. The router then forwards the packet as we described above.

   Activities

See the Activities and Extended Activities section in Unit 8 Lesson 4 in your textbook Introduction to Networking to test what you have learned so far.

[ Previous Section ] [ Unit Contents ] [ Summary ]