About Interdomain Internet Routing

In the lecture, Interdomain Internet Routing by H. Balakrishnan and N. Feamster it is explained how routing between different administrative domains work in the Internet. It is stated that the Internet service is provided by a large number of commercial enterprises called autonomous systems (ASes) in competition of each other but are being required to cooperate for global connectivity. These ASes implements some sets of policies in deciding how to route its packets to the rest of the Internet and how to export the routes of itself, its customers and learned routes to other ASes. In the first part of the lecture the Inter-AS business relationships, transit and peering were discussed. Transit is a provider-customer relationship wherein the provider charges its customers for Internet access. Meanwhile peering on the other hand is a mutual agreement between two providers to allow access to subset of each other's routing tables wherein a financial settlement may not be involve as long as the traffic ration is not highly symmetric. Although peering is beneficial to ASses, its main drawback would be it not being able to give revenue, transit relationships does. Internet Service Providers (ISPs) filter the routes to export and most of them end up providing selective transit. While in importing, routes are imported in the following order: customer, peer and provider.

In the second part of the lecture, the most important features of the Border Gateway Protocol (BGP) were discussed. The BGP was designed to meet three important needs scalability, policy and cooperation under competitive circumstances. This is so that the Internet routing infrastructure remained scalable as the number of connected networks increased, the ASses are capable of implementing various forms of routing policy and to gracefully handle the transfer the "backbone" Internet infrastructure from single to multiple administrative entities. BGP runs on TCP over port 179. The two messages sent by BGP routers are UPDATE messages which announce a change or removal of route and the KEEPALIVE messages to ensure that it is still functioning properly. BGP disseminates routes within between routers on the different ASses using eBGP sessions and iBGP sessions for routers on the same AS. BGP allows the policy expression of ASses by allowing network operators to configure routers to manipulate route attributes when disseminating routes. The most important attributes are LOCAL PREF, ASPATH and MED arranged in priority order. In the end, the authors concluded that although BGP is a simple protocol, its operation in practice is extremely complex because of its configuration flexibility.

This lecture is a great introduction on how routing in the Internet happens between its domains. The design of the internet prioritized the needs of the military, on the other hand interdomain routing prioritize the need of the many administrative systems to be truly independent of each other but still cooperate. BGP gives the ASses the ability to implement their own policies based on their best interest. As consequence of this, the consumers that rely on these providers may not be getting the best possible service.