Networks Horizon

Continue....
Border Gateway Protocol 
BGP-version 4- Part 3



BGP Neighbor Configuration.


Start BGP Neighborship


Unlike all other routing protocols, BGP neighborship is unicast by default. Other difference is that BGP keeps neighborship establishipment separate from actual route advertisement. In other routing protocols (IGPs), there is no need to statically configure the neighbors. It means, BGP neighbors are not discovered automatically, they are statically configured. This is why it is called a connection oriented protocol.


R1(config_router)#neighbor <neigbor IP address> remote-as <neighbor AS number>


There are two ways to get subnets into BGP

• With "network" Command
• With Redistribution

BGP Network Command (Subnet advertisement)


network <subnet> mask <mask> ===>>  classless
network <subnet/ip address>  ====>>classful


If you leave the mask keyword out it uses the default.


Take into consideration that the network command reveals the actual subnet to advertise, not the interface participating in the routing process unlike all other routing protocols.




Redistribution in BGP


Use the redistribute command to redistribute routes from one routing domain to another.
The redistribute command is used to inject routes from one routing domain into another routing domain. You can conditionally control the redistribution of routes between routing domains by defining route maps


Q.Can "redistribute" command redistribute the loopback interface/addresses?

• redistribute connected (optional-route map)
• redistribute static (optional-route map)
• redistribute ospf/rip/is-is/eigrp

If you have a need to dynamically advertise directly connected networks, you might choose to "redistribute connected " command to advertise connected networks into the local BGP process, which will limit the amount of static configuration. You could get the same result by using the BGP "network" command, which would require more configuration and dynamic updation of routes.


This is not a recommended practice to advertise IGP routes into BGP via network command because IGP routes tend to change often and you will not want BGP to constantly add, change, or remove IGP redistributed routes on a regular basis. Rather, you can configure BGP to have IGP routes redistributed directly into BGP by using the "redistribute protocol(ospf/rip/eigrp)" command. 


Resetting BGP Neighborship

If you change a BGP filter, weight, distance, version, or timer, or make a similar configuration change, we must reset BGP connections for the configuration change to take effect. Use either of the following commands in EXEC mode to reset BGP connections:

Command	Purpose
clear ip bgp <neighbor-address>	Reset a particular BGP connection.
clear ip bgp *	Reset all BGP connections.

Soft reset 

The traditional method of clear ip bgp * is disruptive. Soft reconfiguration was introduced in Cisco  to facilitate non disruptive changes in BGP.

clear ip bgp {* | address | peer-group-name} soft out

Soft reset allows policies to be configured and activated without clearing the BGP session like clear ip bgp * command.  It is always applied on outgoing routes and does not have any effect on incoming policy change. This command is always preferred because of its non disruptive nature.


If all BGP routers support the route refresh capability, use the clear ip bgp {* | address | peer-group name} in command. You need not use the soft keyword, because soft reset is automatically assumed when the route refresh capability is supported.


For example::

The following example clears all entries from the host name-and-address cache (hard reset):

clear host *

The following example clears the inbound session with the neighbor 131.108.1.1 without resetting the session:

clear ip bgp 131.108.1.1 soft in

The following example clears the outbound session with the peer-group corp without resetting the session:

clear ip bgp corp soft out

                                  Please read BGP-version 4- Part 4 for more information.