Embedded Web Server for the CR16
incremental window movements resulting in extremely poor TCP performance. It occurs when a
sending TCP foolishly decides to send tiny data segments because the receiving TCP advances
its Window by a tiny amount, although both sender and receiver have a large total buffer space
available. Subsequent studies led to the development and standardization of both sender and
receiver algorithms to preclude it (for those interested, see RFC 1122: 18.104.22.168 and 22.214.171.124).
The Internet Protocol puts the IP in TCP/IP. It is TCP/IPs Network protocol. IP comprises two
basic functions: addressing and fragmentation. Just like TCP, IP encapsulates its data by
prepending it with a header as illustrated in Figure 2.
Its easy to get confused as to just why we need an IP address in the first place. If your PC
sits on an Ethernet or other LAN (Local Area Network), isnt its MAC address unique? Why not
simply use this address instead of requiring yet another one?
The answer is straightforward. Remember that the Internet is not simply one big LAN; rather
the Internet is defined as a network of networks, or perhaps better stated, a network of LANs.
If everybody were a node on one great big homogenous network, all running the same Link layer
protocol (Ethernet, for example), there would be no need for a separate addressing scheme.
The fact is however, that many disparate networks exist, all operating incompatible Link layer
Time to Live
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3
Figure 2. IP Header Format
The MAC Address:
Every host on a LAN is uniquely identified at the Data Link layer by its Link layer, or MAC
address (Media Access Control). Neighboring nodes on any given LAN communicate with each
other based on this physical address. However, a node on an Ethernet cannot directly
communicate with a node on a Token Ring network and vice versa. Likewise, nodes speaking