Network Complexity

In this post, I'll talk about my experience about the Network Complexity.

I started working on this topic in 2010 during my internship at Cisco Systems.

During those 10 last years, there is an increase of the global traffic on Internet, the number of prefixes announced on the Internet reached more than 300000 prefixes in 2010.

Nombre de préfixes annoncés sur l'Internet - http://upload.wikimedia.org/wikipedia/commons/thumb/c/ce/BGP_Table_growth.svg/600px-BGP_Table_growth.svg.png

Because there will be more and more users of the Internet, there will be more and more prefixes. It is not possible to stop the increase of the number of prefixes, but is there a way to contain or to control this increase?

A network architect has to choose between different ways to obtain a network, but which way is the most suitable based on the needs? For example, it is useless to use BGP routing protocol to allow the communication of 2 networks, when you can use static routing.

A research group test a new BGP attribute [1], the problem is that some equipments of the global Internet can't support this attribute. The result is that those equipments reset their BGP peering, and it has a global impact. How did a local value can affect the global Internet traffic?

A research group at the IRTF on the network complexity has been created. The aim of this group is described in details in the charter of the group [2].

To follow the work progress : http://networkcomplexity.org/

References
[1] E. Romijn, “RIPE NCC and Duke University BGP Experiment” 31 August 2010. Available at http://labs.ripe.net/Members/erik/ripe-ncc-and-duke-university-bgp-experiment
[2] IRTF Network Complexity Research Group NCRG. Available at http://irtf.org/ncrg