Week 2 | Review on Interdomain Internet Routing by Hari Balakrishnan
The internet is such a vast architecture whose view of its internal workings have already been simplified to us, users. For example, as you are reading this, dear reader, hundreds of packets are travelling hundreds of miles around the world, passing through routers and getting carried through by links. The paper Interdomain Internet Routing discusses what’s happening behind the scenes of the internet – how your requested content from the other side of the world arrives to you in a split second!
This paper is a really nice read after having read the past 4 papers* that discussed how distributed (rather than a single centralized) administration was one of the original goals of the internet and is fortunately still being preserved through the years. Distributed management allows administrative borders possible between networks which then allows these networks to have their own rules and policies governing their respective spaces. This division also calls for a system with a set of rules and protocols to manage the exchange of messages between each network.
This paper discussed a lot about addresses and how they allow delivery of messages to their intended destinations. The paper also described topological addressing (use of prefixing such that addresses could be chopped to segments to determine to which intermediary router it should be sent first) which allowed communication and routing between millions and billions of network possible.
It was also interesting to know that there exists a “competitive cooperation” between different domains (a.ka. Autonomous Systems) in the pursuit of profit and interconnection. Despite competition, there is still the need to cooperate and communicate which forced them to make compromises and still find a way to become profitable.
Two kinds of relationships that exist between networks were then discussed. These are transit and peering. It’s just notable that terms have actually developed to describe these kinds of relationships which just shows that there was great necessity to define these relationships especially in contracts and financial settlements. With these relationships, the paper have also expounded on the different instances by which ISPs would prefer to prioritize one request over another in the pursuit of making money, saving money, and pursuing their own benefit.
The Border Gateway Protocol (BGP), which allows interconnection between different autonomous systems, were also widely discussed. It is interesting to also know that there actually exists two types, eBGP (external) and iBGP (internal). The paper made a good point to clear out that iBGP is not the same with Internal Gateway Protocols (protocols which are responsible for path metrics inside the same network). And these two types of BGP (whether internal or external) are still the ones responsible in knowing the addresses of the external domains they’ll be connecting on. Different attributes such as NEXT HOP, A SPATH, LOCAL PREF, and MULTIPLE-EXIT DISCRIMINATOR, were also discussed to give a view on how networks prioritize and determine as to which router it will route the message next.
Overall, the paper provided a great overview on how different domains allow routing to and from them and the many factors that affect their prioritization and willingness to do so. The paper also never failed to give an insight to the current situation and scenario, the problems that have arose and the solutions applied. Mistakes and intentional attacks are not foreign in the Interne such as the real life example of Youtube being inaccessible to consumers due to a single person’s error and the lack of control and checks in succeeding entities where the error got passed on. This just emphasises the fact that at the end of the day, we are still interconnected and one’s error and wrongdoing could affect the whole world.
On a one last note, the line “… good technical research can be be shaped and challenged by commercial realities” hit me the most and made me realize that each domain in the network (and even in life) have their own motives and agendas and these affect their decisions in different scenarios. As what the past papers* said, investment and funding affect the direction as to which network development is going.
P.S. I just remembered that this paper could also be related with the current Internet scene here in the Philippines – in the news, many groups have already expressed concern regarding our Internet connectivity which is super slow and whose service quality is below par yet but whose prices are already skyrocketing.
Photo from ASEAN DNA.
Some people have attributed this problem with ISP monopoly where, according to them, many ISPs are actually owned by only 3-4 business conglomerates. Despite these, it is still heartwarming to know that there are efforts in making the Internet available to everyone here in the country, such as WifiNation and Internet.org’s launch here in the Philippines.
Another P.S. (Aug 22, 2015, 12:01 AM): Just came across a Youtube link to the live streaming (c/o Rappler) of the Senate Hearing on PH Internet speed last August 18, 2015.