Summary
The lecture entitled “Interdomain Internet Routing”, from the name itself explains how routing between different administrative domains works in the Internet. An important issue in the current Internet routing layer is tackled initially which pertains to the difficulty of global connectivity of various end-hosts in the Internet. This is due to unequal creation of Internet service providers (ISPs) which are categorized according to the size of their network. Meanwhile, a true picture of the Internet routing is introduced. The existing Internet routing infrastructure is divided into autonomous systems (ASes) that exchange reachability information using the current wide-area routing protocol called BGP (Border Gateway Protocol).
The next part discusses two prevalent forms of relationships happening between different ASes namely transit and peering. The former which involves a financial settlement, requires an ISP to provide access to all (or most) destinations in its routing tables. The latter allows two ASes to provide mutual access to a subset of each others' routing tables and may not involve any financial transactions. ASes involved in either transit or peering needs to construct export policies (rules on what routes to export) and import policies (rules on what routes to install) so as not to lose revenue.
The third section is dedicated to BGP which includes the protocol's design goals, its attributes, and how it disseminates routes within an AS. The current structure of BGP was influenced by these 3 factors: scalability (the Internet routing infrastructure must be scalable as the size of the network increases); policy (AS must enforce various routing policies); and cooperation under competitive circumstances. A BGP consists of several attributes such as NEXT HOP (IP address of the next-hop router), ASPATH (sequence of AS identifiers that the route has traversed), LOCAL PREF (first attribute considered in selecting routes) , and MED (multiple-exit descriminator). Among these, the last three are the most important. Disseminating routes within an AS are done using two types of BSP sessions: eBGP are between BGP in different ASes while iBGP are between BGP in the same AS. The dissemination is done with care using two important goals: loop-free forwarding and complete visibility.
Lastly, failures over BGPs are discussed, including its causes and solutions that has already been made. Moreover, scalability is further explained with the discussion of multi-homing.
Evaluation
This lecture provides a good background on the technicalities behind interdomain Internet routing. The authors have done a great job of including a detailed discussion of BGP’s structure and the goals that served as bases for the protocol’s creation.
Open issues regarding interdomain Internet routing have also been recently identified through surveys and studies done by a group of researchers [1]. Some of these include slow convergence and chattiness of BGP, scalability problems due to multi-homing which is discussed in the paper, security issues, etc. They searched for the possible causes of these issues and if there are any studies done to resolve these issues.
Reference:
[1] M. Yannuzzi, X. Masip-Bruin, and O. Bonaventure, Open Issues in Interdomain Routing: A Survey, IEEE Network, November/December 2005.
The lecture entitled “Interdomain Internet Routing”, from the name itself explains how routing between different administrative domains works in the Internet. An important issue in the current Internet routing layer is tackled initially which pertains to the difficulty of global connectivity of various end-hosts in the Internet. This is due to unequal creation of Internet service providers (ISPs) which are categorized according to the size of their network. Meanwhile, a true picture of the Internet routing is introduced. The existing Internet routing infrastructure is divided into autonomous systems (ASes) that exchange reachability information using the current wide-area routing protocol called BGP (Border Gateway Protocol).
The next part discusses two prevalent forms of relationships happening between different ASes namely transit and peering. The former which involves a financial settlement, requires an ISP to provide access to all (or most) destinations in its routing tables. The latter allows two ASes to provide mutual access to a subset of each others' routing tables and may not involve any financial transactions. ASes involved in either transit or peering needs to construct export policies (rules on what routes to export) and import policies (rules on what routes to install) so as not to lose revenue.
The third section is dedicated to BGP which includes the protocol's design goals, its attributes, and how it disseminates routes within an AS. The current structure of BGP was influenced by these 3 factors: scalability (the Internet routing infrastructure must be scalable as the size of the network increases); policy (AS must enforce various routing policies); and cooperation under competitive circumstances. A BGP consists of several attributes such as NEXT HOP (IP address of the next-hop router), ASPATH (sequence of AS identifiers that the route has traversed), LOCAL PREF (first attribute considered in selecting routes) , and MED (multiple-exit descriminator). Among these, the last three are the most important. Disseminating routes within an AS are done using two types of BSP sessions: eBGP are between BGP in different ASes while iBGP are between BGP in the same AS. The dissemination is done with care using two important goals: loop-free forwarding and complete visibility.
Lastly, failures over BGPs are discussed, including its causes and solutions that has already been made. Moreover, scalability is further explained with the discussion of multi-homing.
Evaluation
This lecture provides a good background on the technicalities behind interdomain Internet routing. The authors have done a great job of including a detailed discussion of BGP’s structure and the goals that served as bases for the protocol’s creation.
Open issues regarding interdomain Internet routing have also been recently identified through surveys and studies done by a group of researchers [1]. Some of these include slow convergence and chattiness of BGP, scalability problems due to multi-homing which is discussed in the paper, security issues, etc. They searched for the possible causes of these issues and if there are any studies done to resolve these issues.
Reference:
[1] M. Yannuzzi, X. Masip-Bruin, and O. Bonaventure, Open Issues in Interdomain Routing: A Survey, IEEE Network, November/December 2005.
No comments:
Post a Comment