How to Make Autonomous System (AS) Connections
Interior Gateway Protocols like OSPF and IS-IS enable you to set up networks and exchange routing information within your network. This kind of network has been aptly labeled an autonomous system (AS). An AS is a set of routers and devices, or even a set of networks, controlled by a single entity.
If you have any need to access the Internet, either to grab information or to use it as a transport to other networks, you must be able to connect outside the AS. These connections are established using peering relationships, where one AS connects to another (a peer) using Border Gateway Protocol (BGP).
To connect ASs to each other and establish a peering relationship, you must configure BGP on both peering routers.
Whereas an IGP like OSPF is easy to configure and works on its own once you enable it, you must explicitly configure BGP. BGP can be rather unwieldy at times, primarily because when you’re using BGP, you’re connecting to a router outside your control. Therefore, you’re likely to want stricter security in terms of the information you make available to your peers as well as what they send to you.
Imagine that you have a simple network with two ASs, each of which has a gateway router. You want to connect the two networks using BGP. To establish a connection, BGP requires a little bit of information:
You must identify the AS to which each of the peering routers belongs. Every AS in the world is uniquely identified by an AS number. These numbers are handed out by the Internet Assigned Numbers Authority (IANA) and are used to specify not only the peering router but also the peering AS for each BGP session.
You must decide on a group for the peering session. BGP groups everything so that you can have logical sets of connections that all behave more or less the same way.
Imagine, for example, that you have multiple connections between your network and a neighbor network. You may have all the same configuration on these links except that they are between different routers (to provide a redundant link). To simplify the configuration, you group them and call them collectively “Those guys.” All configuration for “Those guys” is employed on each individual session within the group.
You must know the specific IP address of the interface to which you’re connecting. This address is the neighbor address because it’s the neighboring interface with which you are peering.
The need for the specific IP address is the reason why BGP is an EGP and not an IGP. While you can use BGP to interconnect all of the routers within your network, the fact that you have to explicitly configure each connection can be a pain. It is far simpler to use a lighter-weight protocol like OSPF and save the heavy-duty protocols for the connections outside your network.
As a general rule and best practice, you want your IGP to carry local and interface routes. You want to leave the heavy lifting for BGP. BGP was built to handle large numbers of routes. IGPs, on the other hand, were designed to reconverge as fast as possible in the event of a failure (link, router, or other type of failure).
Keep in mind that the other guy must have all the same information as well. For this session to work, both of you have to explicitly configure BGP to each other.