Information Technology Reference
In-Depth Information
The packet-forwarding database is built based on the information contained in the IP RIB
and IP Address Resolution Protocol (ARP) table. A prefix lookup in the RIB is performed
to determine the next-hop address and the outgoing interface. The actual Layer 2 header is
built based on the information from the IP ARP table. Frame Relay and ATM maps are other
examples used to map Layer 3 addresses to Layer 2 addresses. Two general types of RIB
lookup operations are supported in Cisco IOS software:
•
Classless
—The longest matching prefix is looked up. If no matching prefixes are
found, the default route, if present, is used. IP classless lookup has been the default
(although it is still shown in the running configuration) since Cisco IOS software
Release 11.3.
•
Classful
—Longest-match lookup. Supernets and the default route are not considered
if the routing table contains a subnet of the destination major network (the classful
network of the address being resolved).
Switching Paths
Within Cisco IOS software, three general switching paths are supported, dependent on the
hardware platforms and configurations:
•
Process switching
•
Cache-based switching
•
Cisco Express Forwarding (CEF)
The next sections discuss each switching path in greater detail.
Process Switching
Process switching is the most basic form of switching and is universally available on all
Cisco routers.
Process switching
refers to the fact that the CPU is directly involved in the
process required to forward the packet. The packet is switched at the process level within
IOS. In other words, the forwarding decision is made by a process scheduled by the IOS
scheduler and running as a peer to other processes on the router, such as routing protocols.
Processes that normally run on the router are not interrupted to process-switch a packet. For
IP packets, the forwarding process is IP Input.
Figure 2-3 shows the main components of typical IP process switching. The following list
outlines the process:
An IP packet received from the inbound interface is queued in the Synchronous
Dynamic RAM (SDRAM) packet memory.
1
The processor copies the packet to the system buffer area in Dynamic RAM (DRAM),
where the IP Input process begins its Layer 3 and Layer 2 processing of the packet.
2
