Netgear XCM8806 - 8800 SERIES 6-SLOT CHASSIS SWITCH ユーザーズマニュアル

Origin code

Multi Exit Discriminator (MED)

IGP distance to the next hop

Source session (EBGP or IBGP) 

ECMP does not install an additional path if the next hop is the same 

as that of the best path. All paths within a multipath must have a 
unique next hop value.

BGP ECMP does not affect the best path selection. For example, the router continues to 
designate one of the paths as the best path and advertise this best path to its neighbors. 
EBGP paths are preferred to IBGP paths.

The BGP ECMP feature allows you to define the maximum number of equal cost paths (up to 
eight) in a multipath. A multipath for an IBGP destination is called an IBGP multipath, and the 
multipath for an EBGP destination is called an EBGP multipath. 

If there are more equal cost paths for a destination than the configured maximum, the BGP 
identifier for the advertising BGP speaker is used to establish a path priority. The lower BGP 
identifier values have priority over the higher values. For example, if the configuration 
supports 4 paths in a multipath, only the four paths with the lowest BGP identifier values 
become part of the multipath.

To enable or disable BGP ECMP, enter the following command:

configure bgp maximum-paths <max-paths>

The 

max-paths

 setting applies to BGP on the current VR. Specify more than 1 path to enable 

BGP ECMP and define the maximum number of paths for IBGP and EBGP multipaths. 
Specify 1 path to disable ECMP. To display BGP ECMP configuration information, use the 

show bgp

 command.

BGP Static Network

XCM8800 BGP allows users to add static networks in BGP, which will be redistributed 
(advertised) into the BGP domain if there is a corresponding active route in the IP routing 
table. Users can associate a policy with the static BGP network to change or to set the route 
attributes before the route is advertised to the BGP neighbors. 

To create a static BGP network, use the following command:

configure bgp add network {address-family [ipv4-unicast | ipv4-multicast]} 

<ipaddr>/<mask_len> {network-policy <policy>}