Border Gateway Protocol (BGP) Routing Protocol Overview
Border Gateway Protocol (BGP) is an External Gateway Protocol (EGP), which means it is designed to act as a routing protocol on the edge of your network, passing routing information about the structure of your network behind your gateway router.
This information can then be sent on to other BGP routers, informing them which networks are found behind the BGP router. A BGP router announces routes that they have learned and can also retransmit routes learned from the IGPs found on their networks, as illustrated in the following figure.
BGP's purpose is not only to exchange its information, but also to exchange network reachability and availability information for the Autonomous Systems (AS) paths with other BGP systems on the network. This process allows all systems to construct topology graphs of the entire network infrastructure on both sides of the BGP link.
This process also allows these systems to identify loops and other issues that may affect network performance and availability.
BGP has several versions of BGP, and version 4 is the current one. When two BGP systems start communicating, they attempt to use version 4 of the protocol. If one system does not support version 4, they negotiate down to older versions of the protocol until they find a version they both can use.
As with other routing protocols, BGP supports modification of the preferred route through modification of metric values. BGP does so with weights. A weight is the administrative value assigned to the local router. This value can be anything between 0 and 65,535 with a default value on Cisco routers of 32,768.
To force traffic through a specific router, you can have its neighbors assign higher weights to routes than they have learned from that router. So, higher weights identify a preferred route, whereas with other routing protocols lower costs identify the preferred route.