to determine the remote computer’s address before a connection can be made. A designated computer on the network will contain a database of computer names and their corresponding IP and MAC addresses. The user’s computer will send a query to the name database computer, and the database computer will send a response. Both the query and the response are very short. There is no need to divide the query or response between multiple packets, so the complexity of TCP is not required. If there is no response to the query after a period of time, the query can simply be resent.
The User Datagram Protocol (UDP) is designed for communications that do not require division among multiple packets and subsequent reassembly. UDP does not keep track of what is sent.
UDP uses port numbers in a way that is directly analogous to TCP. There are
Figure 5-16. Ethernet Packet Header
The UDP header is shorter than a TCP header. UDP also uses a checksum to verify that data is received uncorrupted.
The Internet Control Message Protocol (ICMP) is also a simplified protocol used for error messages and messages used by TCP/IP. ICMP, like UDP, processes messages that will fit into a single packet. ICMP does not, however use ports because its messages are processed by the network software.
The Domain Name System
Computer users usually prefer to use text names for computers they may want to open a connection with. Computers themselves, require 32 bit IP addresses. Somewhere, a database of network devices’ text names and their corresponding IP addresses must be maintained.
The Domain Name System (DNS) is used to map names to IP addresses throughout the Internet and has been adapted for use within intranets.
For two DNS servers to communicate across different subnets, the DNS Relay of the
Mapping Domain Names to Addresses
The Domain Name System (DNS) servers are organized in a somewhat hierarchical fashion. A single server often holds names for a single network, which is connected to a root DNS server – usually maintained by an ISP.
54
