Internet Connectivity Research
   Research into the connectivity of the Internet has been going on for quite a few years. The topic is made very interesting by the fact that no one central authority controls Internet routing or connectivity. Thus, the Internet grows through cooperative means not through permisson or control of a central authority.
  Mapping connectivity of the Internet requires network connectivity data collection. Traditionally, this has been done using ping-like packets with low Time To Live (TTL) field values which are increased. The low TTL value generally results in a ping packet "expiring" before it reaches its destination. When a packet's TTL reaches zero, the router processing the packet may generate a ICMP packet reporting the TTL expiration and send it back to the ping packet originator. Included in the ICMP packet is "a" source address of the router that detected the TTL expiration. (This is a well known technique that is used for the traceroute tool.)
  While this is an exciting way to check the connectivity of the Internet, is has a few issues.

1. In order to work, it relies on routers actually responding to TTL expiration with an ICMP error message. This does not always happen.
2. It is an active method of data collection. In other words, network connectivity mapping using the traceroute technique requires that packets be sent all over the Internet. It is possible that network administrators might block the incoming packets and effectively block the network connectivity mapping efforts.
3. Point of view matters. A system mapping the Internet from the AT&T network will see a different connectivity map than system mapping the network from a Comcast network.
4. The Internet is just too large to map using a traceroute style of mapping.

 

  Sites of Interest for other Internet Connectivity Research

Potential solution

To address some of these network mapping issues, CyanLine has performed research into mapping the Internet using BGP feeds and BGP connectivity. (Using data collected by the Oregon Routeviews project and processed by CAIDA.) While we are working on a paper to detail the findings, we have been excited by the results to date. We are releasing the images that we have made from the BGP data and look forward to your comments. 

. . . . . . . . . . . . . . . . . .