Users Guide

1116 IP Routing
IP Path MTU and Path MTU Discovery
The IP stack maintains an IP MTU for each route in its routing table.
Conceptually, the route’s path MTU defaults to the IP MTU of the outgoing
interface. The IP MTU of an interface is set automatically based upon the
switch MTU. If the switch receives an ICMPv4 Fragmentation Needed or
ICMPv6 Packet Too Big message, the IP stack sets the corresponding route’s
path MTU to the value in the ICMP message as long as it is less than the
switch MTU minus the Ethernet frame header length.
RFC 1191 explains how a router can initiate IPv4 path MTU discovery. The
basic idea is that the router sends IPv4 packets with the “don’t fragment” bit
set. The router initially assumes that the path MTU is equal to the IP MTU
of the outgoing interface. If the packet is too big to reach its destination
without fragmentation, a router in the path will return a packet too big
message. The originator reduces its estimate of the path MTU and continues
the process until packets reach the final destination.
Path MTU discovery is required for IPv6. In IPv6, only the originator is
allowed to fragment. Any packet too large to reach its destination triggers a
packet too big message, updating the IP stack’s path MTU for the
destination.
Route Preference Values Preference values are as follows:
• Local—0
• Static—1
• OSPF Intra—110
• OSPF Inter—110
• OSPF External—110
• RIP—120
Table 29-2. IP Routing Defaults (Continued)
Parameter Default Value