Release 15.5SY Supervisor Engine 6T Software Configuration Guide

Hardware Layer 3 Switching

Hardware Layer 3 switching allows the PFC and DFCs, instead of the RP, to forward IP unicast traffic between subnets. Hardware Layer 3 switching provides wire-speed forwarding on the PFC and DFCs, instead of in software on the RP. Hardware Layer 3 switching requires minimal support from the RP. The RP routes any traffic that cannot be hardware Layer 3 switched.

Hardware Layer 3 switching supports the routing protocols configured on the RP. Hardware Layer 3 switching does not replace the routing protocols configured on the RP.

Hardware Layer 3 switching runs equally on the PF3 and DFCs to provide IP unicast Layer 3 switching locally on each module. Hardware Layer 3 switching provides the following functions:

• Hardware access control list (ACL) switching for policy-based routing (PBR)
• Hardware flow-based switching for TCP intercept and reflexive ACL forwarding decisions
• Hardware Cisco Express Forwarding (CEF) switching for all other IP unicast traffic

Hardware Layer 3 switching on the PFC supports modules that do not have a DFC. The RP forwards traffic that cannot be Layer 3 switched.

Traffic is hardware Layer 3 switched after being processed by access lists and quality of service (QoS).

Hardware Layer 3 switching makes a forwarding decision locally on the ingress-port module for each packet and sends the rewrite information for each packet to the egress port, where the rewrite occurs when the packet is transmitted from the switch.

Hardware Layer 3 switching generates flow statistics for Layer 3-switched traffic. Hardware Layer 3 flow statistics can be used for NetFlow. (See Chapter 15, “NetFlow Hardware Support”.)

Layer 3-Switched Packet Rewrite

When a packet is Layer 3 switched from a source in one subnet to a destination in another subnet, the switch performs a packet rewrite at the egress port based on information learned from the RP so that the packets appear to have been routed by the RP.

Packet rewrite alters five fields:

• Layer 2 (MAC) destination address
• Layer 2 (MAC) source address
• Layer 3 IP Time to Live (TTL)
• Layer 3 checksum
• Layer 2 (MAC) checksum (also called the frame checksum or FCS)

Note Packets are rewritten with the encapsulation appropriate for the next-hop subnet.

If Source A and Destination B are in different subnets and Source A sends a packet to the RP to be routed to Destination B, the switch recognizes that the packet was sent to the Layer 2 (MAC) address of the RP.

To perform Layer 3 switching, the switch rewrites the Layer 2 frame header, changing the Layer 2 destination address to the Layer 2 address of Destination B and the Layer 2 source address to the Layer 2 address of the RP. The Layer 3 addresses remain the same.

In IP unicast and IP multicast traffic, the switch decrements the Layer 3 TTL value by 1 and recomputes the Layer 3 packet checksum. The switch recomputes the Layer 2 frame checksum and forwards (or, for multicast packets, replicates as necessary) the rewritten packet to Destination B’s subnet.

A received IP unicast packet is formatted (conceptually) as follows:

After the switch rewrites an IP unicast packet, it is formatted (conceptually) as follows:

Hardware Layer 3 Switching Examples

Figure 33-1 shows a simple network topology. In this example, Host A is on the Sales VLAN (IP subnet 171.59.1.0), Host B is on the Marketing VLAN (IP subnet 171.59.3.0), and Host C is on the Engineering VLAN (IP subnet 171.59.2.0).

When Host A initiates an HTTP file transfer to Host C, Hardware Layer 3 switching uses the information in the local forwarding information base (FIB) and adjacency table to forward packets from Host A to Host C.

Figure 33-1 Hardware Layer 3 Switching Example Topology