Troubleshooting guide

BGP Overview

By defining autonomous systems, a demarcation point is created between organizations and the Internet.

Within the AS, information about all networks can be transmitted to every other router using some type of

IGP, such as OSPF or RIP. However, since the Internet is so vast, it would be impractical for all routers to

hold routes to all networks in all of the autonomous systems. The capability of the Internet to identify an

entire organization by means of a unique integer allows a large reduction in the amount of information that

needs to be held in routing tables or transmitted in routing updates between autonomous systems. The

result is a considerably smaller amount of summarized information exchanged using BGP between

different autonomous systems. This level of hierarchy is essential to the successful operation and

maintenance of the Internet.

Initially, AS numbers were only 16-bit integers. However, as the Internet continues to experience massive

growth, engineers have expanded the AS number space from 2 bytes to 4 bytes, providing over 4 billion

AS numbers. January 2009 marked the transition to allocate 32-bit AS numbers by default.

External BGP (eBGP) and Internal BGP (iBGP)

eBGP uses BGP between peers in different autonomous systems, and iBGP uses BGP between peers

within the same AS.

Do not confuse eBGP with EGP, a nearly obsolete protocol once used on the Internet. Also, do not confuse

iBGP with an IGP, such as OSPF. Service providers use iBGP to distribute BGP routes between routers

within an AS. However, service providers usually still need to run an IGP to generate routes for traffic

within the AS.

eBGP allows an organization to peer and exchange routes with a service provider. MPLS implementations

by a service provider allow the exchange of private subnets between remote sites through the provider’s

network. eBGP is also used between different service providers to facilitate the Internet backbone.

Customers can peer using eBGP with service providers to exchange public IP addresses that they have

purchased from IANA or to exchange private routes to other remote sites over an MPLS circuit.

iBGP is used between BGP routers within the same AS. iBGP routers prevent routing loops by following a

rule where no updates learned from internal peers are sent to other internal peers. This means that iBGP

routers will only propagate a route if the route originates in the transmitting router or if the route is directly

connected to the transmitting router. As a result, iBGP routers must be fully meshed to have a complete

knowledge of the network.

Route reflectors are sometimes identified and used in an iBGP network to reduce the number of peering

sessions required.

As of release 18.1, AOS supports 32-bit AS numbers.

AOS devices have a 16 BGP neighbor limit. Since iBGP routers must be fully meshed, AOS

devices are only suitable for small scale iBGP applications.

AOS devices cannot act as a route reflector for iBGP setups. AOS devices can peer with a

route reflector acting as a client, however, the connection is no different than a regular

iBGP peer.