ABSTRACT
Network Address Port Translation (NAT/NAPT) ◾ Understand the protocols for dynamically opening holes with NAPT for application protocols ◾ Learn the diagnostic operation of the Internet Control Message Protocol (ICMP) ◾ Understand the salient operations of mobile IP
10.1.1 THE NEED FOR NETWORK AND LINK LAYERS
As a way of introducing this subject, let us ask the question, why do we need both the network and link layers? Why not just have one or the other? The answers to these questions involve three issues: (1) performance, (2) security and (3) cost. The switch cost is affected by the size of memory required for the switch table, and the performance depends on the time needed to search the switch table. When there is only one layer, either Network or Link, the size of the switch or routing table must be large enough to contain every host in the Internet, which would render the cost prohibitively high, the table would never converge, and the search time would be forever. When there is only a Link layer and no Network layer, broadcast storms impact performance and security. With a router and layer 2 switch, the network can be split into subnets and address all of these problems. Each layer 2 switch only handles the switching of a subnet and each router only handles the switching among subnets. In this case, the sizes of the layer 2 switch table and layer 3 routing table, as well as the associated search time, are in the usable range. This separation of link and network layers also reduces the cost. “Divide and conquer” is the methodology for handling packet switching. In order to ensure that the size of the routing tables is reasonable, the routers are separated into two categories: interior gateway and exterior gateway. The former is inside a domain while the latter is outside. Interior gateway routing is performed by campus class routers, while exterior gateway routing is accomplished by carrier class routers, as shown in Figure 10.1.
