When routing full Internet tables which can hold more than 90,000 routes you need to rely on a very robust and highly scalable routing protocol. BGP (Border Gateway Protocol) is an interautonomous system routing protocol. An autonomous system would share common administration and routing policies inside of a network or group of networks. BGP is becoming the standard routing protocol for ISPs (Internet Services Providers) replacing the older EGP (Exterior Gateway Protocol). ISPs use BGP to route between other providers and or customer networks also running BGP. Many times ISP customers with use an IGP (Interior Gateway Protocol) to route their interior networks and will also sometimes use it to exchange routes with their ISP. If BGP is used inside of an AS (autonomous system) its referred to as IBGP (Interior Border Gateway Protocol) but when its used to route between AS (autonomous systems) its referred to as EBGP (External Border Gateway Protocol).

In order to handle large Internet routing tables BGP must use attributes in order to maintain stability and define routing policies. BGP also makes use of CIDR (classless interdomain routing) in order to reduce the size of the routing tables by using Supernets. By using Supernets many routes can be summerized into larger route blocks shaving down the size of the routing updates and routing tables tremendously.

BGP uses TCP to establish communications between BGP peers or neighbors. Once the neighbor relationship is fully established complete routing tables are exchanged the first time. From then on BGP doesn't send out periodic updates, instead when a change is detected in the routing table the BGP neighbors will only exchange the information that has changed. Only the optimal routing path in the BGP table is exchanged.

In order to determine the "optimal path" to a destination when there are multiple paths available BGP uses attributes. In order to design an optimal BGP network you must understand the BGP attributes and how they can be used to effect path selection within the network.

Below are a list of the BGP attributes in order of preference in the BGP path selection process:

• Weight
• Local preference
• Multi-exit discriminator
• Origin
• AS_path
• Next hop
• Community

If a BGP peer receives multiple advertisements to the same destination it must go through the below process based on the attributes above prior to inserting the proper route in its routing tables and propogating it to its neighbors.

• The path specifies a next hop that is inaccessible. > Drop the invalid update.
• Path with the largest weight is preferred.
• Weights are the same. > The path with the largest local preference is preferred.
• Local preferences are the same. > The path that was originated by BGP running on this router is preferred.
• No route was originated. > The route that has the shortest AS_Path is preferred.
• All paths have the same AS_Path length. > The path with the lowest origin type (where IGP is lower than EGP, and EGP is lower than incomplete) is preferred.
• The origin codes are the same. The path with the lowest MED attribute is preferred.
• The paths have the same MED. > The external path is preferred over the internal path.
• The paths are still the same. > the path through the closest IGP neighbor is preferred.
• The BGP router with the lowest IP address specified for the router ID is preferred.

This is an overview of the basic operation of BGP (Border Gateway Protocol), I hope you enjoyed it.