39-12
Cisco Catalyst Switch Module 3110 and 3012 for IBM BladeCenter Software Configuration Guide
OL-12189-01
Chapter 39 Configuring IPv6 Host Functions and Unicast Routing
Understanding IPv6
Limitations
This section applies only to the Catalyst Switch Module 3110.
Because IPv6 is implemented in hardware in the switch, some limitations occur due to the use of IPv6
compressed addresses in the hardware memory. These hardware limitations result in some loss of
functionality and limits some features.
These are feature limitations.
• Load-balancing using equal cost and unequal cost routes is not supported for IPv6 host routes or for
IPv6 routes with a mask greater than 64 bits.
• The switch cannot forward SNAP-encapsulated IPv6 packets. These packets are corrupted before
being forwarded (bridged or routed) and reach the network as corrupted packets.
Note There is a similar limitation for IPv4 SNAP-encapsulated packets, but the packets are
dropped at the switch and are not forwarded as corrupted packets.
• The switch routes IPv6-to-IPv4 and IPv4-to-IPv6 packets in hardware, but the switch cannot be an
IPv6-to-IPv4 or IPv4-to-IPv6 tunnel endpoint.
• Bridged IPv6 packets with hop-by-hop extension headers are forwarded in software. In IPv4, these
packets are routed in software, but bridged in hardware.
• In addition to the normal SPAN and RSPAN limitations defined in the software configuration guide,
these limitations are specific to IPv6 packets:
–
When you egress RSPAN IPv6-routed packets, the source MAC address in the SPAN output
packet can be corrupted.
–
When you egress RSPAN IPv6-routed packets, the destination MAC address can be corrupted.
Normal traffic is not affected.
• The switch cannot apply QoS classification or policy-based routing on source-routed IPv6 packets
in hardware.
• The switch cannot generate ICMPv6 Packet Too Big messages for multicast packets.
IPv6 and Switch Stacks
This section applies only to the Catalyst Switch Module 3110.
The switch supports IPv6 forwarding across the stack much as it does with IPv4 unicast routing. The
stack master runs the IPv6 unicast routing protocols and computes the routing tables. Using distributed
CEF (dCEF), the stack master downloads the routing table to the stack member switches. The member
switches receive the tables and install IPv6 routes into hardware for hardware forwarding.
Note To route IPv6 packets in a stack, all switches in the stack must be running the advanced IP services
feature set.
If a new switch becomes the stack master, the new master recomputes the IPv6 routing tables and
distributes them to the member switches. While the new stack master is being elected and is resetting,
the switch stack does not forward IPv6 packets. The stack MAC address also changes. When the IPv6
address of the stack is specified with an extended universal identifier (EUI) by using the ipv6 address
