3.1.a Packet handling along the path through a network
A packet refers to an IP packet which has a source and destination IP address in the header. Throughout the network the packet is forwarded based on the information in the Router’s Information Base (RIB) with the longest prefix chosen.
Here is a picture of a Ethernet Frame showing the IP packet component.
3.1.b Forwarding decision based on route lookup
Here is a simple network with multiple paths from R1 => R4 and duplicate IP address (R1/PC1):
A IP packet that traversed this network from R1 => R4 or R1 => PC1 would be forwarded on each router based on the following (on up, up interfaces):
- Most specific (longest) prefix
- Lowest administrative distance
- Lowest metric
Here is R1’s routing table:
As you can see there are multiple choices for a route to 10.234.132.203 (PC1/R4). Issuing the command “show ip route 10.234.132.203” will show the routers choice of next-hop for this address:
Why did R1 choose to route traffic to the /24 prefix rather than the directly connected /8 prefix? As per rule one – the longest prefix wins – despite the very low metric of the connected interface the longest prefix is chosen.
There are however multiple routes to R4 – via R2 or R3.
The route in this instance is via g5/0. Querying OSPF will show why:
OSPF has listed g5/0 as the route to 10.234.132.0/24 as it has the lowest cost (gigabit link to switch). If we shutdown this interface the alternate route will show:
3.1.c Frame rewrite
Frame refers to the layer 2 part of the segment. Let’s trace a ping from R1 => R4:
R1(g5/0) ca01.0e97.008c => R3(g5/0) ca03.0ed1.008c
R3(g4/0) ca03.0ed1.0070 => R4(g4/0) ca04.0fd0.0070
Each time the segment is sent over anther link the Ethernet frame is rewritten with updated layer 2 information while the packet (layer 3) is not modified.