Brocade Communications Systems ICX 7250 series, Configuration Manual

Page: 86 / 483

Support for IPv4 multicast routing over GRE tunnels
PIM-DM and PIM-SM Layer 3 multicast protocols and multicast data traffic are supported over GRE tunnels. When a multicast protocol
is enabled on both ends of a GRE tunnel, multicast packets can be sent from one tunnel endpoint to another. To accomplish this, the
packets are encapsulated using the GRE unicast tunneling mechanism and forwarded like any other IPv4 unicast packet to the
destination endpoint of the tunnel. The router that terminates the tunnel (i.e., the router where the tunnel endpoint is an ingress interface)
de-encapsulates the GRE tunneled packet to retrieve the native multicast data packets. After de-encapsulation, data packets are
forwarded in the direction of its receivers, and control packets may be consumed. This creates a PIM-enabled virtual or logical link
between the two GRE tunnel endpoints.
Strict RPF check for multicast protocols
Brocade software enforces strict Reverse Path Forwarding (RPF) check rules on an (s,g) entry on a GRE tunnel interface. The (s,g) entry
uses the GRE tunnel as an RPF interface. During unicast routing transit, GRE tunnel packets may arrive at different physical interfaces.
The strict RPF check limits GRE PIM tunnel interfaces to accept the (s,g) GRE tunnel traffic.
Configuration considerations for GRE IP tunnels
Before configuring GRE tunnels and tunnel options, consider the configuration notes in this section.
• When GRE is enabled on a Layer 3 switch, the following features are not supported on Virtual Ethernet (VE) ports, VE member
ports (ports that have IP addresses), and GRE tunnel loopback ports:
–
ACL logging
– ACL statistics (also called ACL counting)
– MAC address filters
– IPv6 filters
NOTE
The above features are supported on VLANs that do not have VE ports.
• Whenever multiple IP addresses are configured on a tunnel source, the primary address of the tunnel is always used for forming
the tunnel connections. Therefore, carefully check the configurations when configuring the tunnel destination.
• When a GRE tunnel is configured, you cannot configure the same routing protocol on the tunnel through which you learn the
route to the tunnel destination. For example, if the FastIron learns the tunnel destination route through the OSPF protocol, you
cannot configure the OSPF protocol on the same tunnel and vice-versa. When a tunnel has OSPF configured, the FastIron
cannot learn the tunnel destination route through OSPF. This could cause the system to become unstable.
• The tunnel destination cannot be resolved to the tunnel itself or any other local tunnel. This is called recursive routing. This
scenario would cause the tunnel interface to flap and the Syslog message TUN-RECURSIVE-DOWN to be logged. To resolve
this issue, create a static route for the tunnel destination.
GRE MTU configuration considerations
When jumbo is enabled, the default Ethernet MTU size is 9216 bytes. The maximum Ethernet MTU size is 10218 bytes. The MTU of
the GRE tunnel is compared with the outgoing packet before the packet is encapsulated. After encapsulation, the packet size increases
by 24 bytes. Therefore, when changing the GRE tunnel MTU, set the MTU to at least 24 bytes less than the IP MTU of the outgoing
interface. If the MTU is not set to at least 24 bytes less than the IP MTU, the size of the encapsulated packet will exceed the IP MTU of
the outgoing interface. This will cause the packet to either be sent to the CPU for fragmentation, or the packet will be dropped if the DF
(Do-Not-Fragment) bit is set in the original IP packet, and an ICMP message is sent.
IPv4 point-to-point GRE tunnels
Brocade FastIron Layer 3 Routing Configuration Guide
86 53-1003903-04