BGP4+ route reection
A BGP device can act as a route-reector client or as a route reector. You can congure a BGP peer as a route-reector client from the
device that is going to reect the routes and act as the route reector using the neighbor route-reector-client command.
When there is more than one route reector, they should all belong to the same cluster. By default, the value for cluster-id is used as the
device ID. The device ID can be changed using the cluster-id command.
The route-reector server reects the routes as follows:
• Routes from the client are reected to the client as well as to nonclient peers.
• Routes from nonclient peers are reected only to client peers.
If route-reector clients are connected in a full IBGP mesh, you can disable client-to-client reection on the route reector using the no
client-to-client-reection command.
A BGP device advertises only those routes that are preferred ones and are installed into the Routing Table Manager (RTM). When a route
cannot be installed into the RTM because the routing table is full, the route reector may not reect that route. In cases where the route
reector is not placed directly in the forwarding path, you can congure the route reector to reect routes even though those routes are
not in the RTM using the always-propagate command.
BGP4+ route aggregation
A device can be
congured to aggregate routes in a range of networks into a single IPv6 prex.
By default, a device advertises individual BGP4+ routes for all the networks. The aggregation feature allows you to congure a device to
aggregate routes in a range of networks into a single IPv6 prex. For example, without aggregation, a device will individually advertise
routes for networks 2001:db8:0001:0000::/64, 2001:db8:0002:0000::/64, 2001:db8:0003:0000::/64, and so on. You can
congure the device to send a single, aggregate route for the networks instead so that the aggregate route would be advertised as
2001:db8::/32 to BGP4 neighbors.
BGP4+ multipath
The BGP4+ multipath feature can be used to enable load-balancing across dierent paths.
BGP4+ selects only one best path for each IPv6 prex it receives before installing it in the IP routing table. If you need load-balancing
across dierent paths, you must enable BGP4+ multipath using the maximum-paths command under IPv6 address family conguration
mode.
IBGP paths and EBGP paths can be exclusively selected, or a combination of IBGP and EBGP paths can be selected.
The following attributes of parallel paths must match for them to be considered for multipathing:
• Weight
• Local Preference
• Origin
• AS-Path Length
• MED
• Neighbor AS (EBGP multipath)
• AS-PATH match (for IBGP multipath)
BGP4+ multipath
FastIron Ethernet Switch Layer 3 Routing
53-1003627-04 481
To Next Page
Loading...
Need help?
General
