Chapter 3

3.1	Must the loopback address be 127.0.0.1?
A:	No, any class A address with a network ID of 127 is OK, although most systems use 127.0.0.1.
3.2	Identify the routers in Figure 3.6 with more than two network interfaces.
A:	`kpno` has five interfaces: three point-to-point links and two Ethernets. R10 has four Ethernet interfaces. `gateway` has three interfaces: two point-to-point links and one Ethernet. Finally, `netb` has one Ethernet interface and two point-to-point links.
3.3	What's the difference in the subnet mask for a class A address with 16 bits for the subnet ID and a class B address with 8 bits for the subnet ID?
A:	There's no difference: both have a subnet mask of 255.255.255.0, as does a class C address that is not subnetted .
3.5	Is the subnet mask 255.255.0.255 valid for a class A address?
A:	It's valid and it's called a noncontiguous subnet mask since the 16 bits for the subnet mask are not contiguous. The RFCs, however, recommend against using noncontiguous subnet masks .
3.6	Why do you think the MTU of the loopback interface printed in Section 3.9 is set to 1536?
A:	It's a historical artifact. The value is 1024 + 512 but the MTU values printed include any required headers. Solaris 2.2 sets the MTU of the loopback interface to 8232 (8192 + 40), which allows room for 8192 bytes of user data along with the normal 20-byte IP header and 20-byte TCP header.
3.7	The TCP/IP protocol suite is built on a datagram network technology, the IP layer. Other protocol suites are built on a connection-oriented network technology. Read [Clark 1988] to discover the three advantages the datagram network layer provides.
A:	First, datagrams eliminate the need for connection state in the routers. Second, datagrams provide the basic building block on which unreliable (UDP) and reliable (TCP) transport layers can be built. Third, datagrams represent the minimal network layer assumption, allowing a wide range of data-link layers to be used.