Traceroute Statistics

Summary Description

What is Measured

For the TTM project, we determine the path between two Test Boxes with the Traceroute program. The output of this program shows how data travelled between 2 specific boxes at the time the program was run.

The collection of all traceroutes done around a particular moment in time can be used to describe the connectivity of the Internet as a whole. The Internet can be seen as a sparsely connected collection of nodes, where each node is one IP address (and thus, one interface on a router). Sparsely means that it is possible to travel from each node to each other node, but that one may have to travel through intermediate nodes. The properties of this graph are described by the average and variance of the number of nodes that one has to pass to travel from any one node to another: the lower the average, the better the Internet is connected. A lower variance suggest that connectivity to all sites passes the same number of hops, and thus that delays will be more predictable.

By assigning the IP addresses of the traceroute to an AS number, one can see how many AS's were traversed between 2 boxes. As was the case at the AS level, the graph again shows how well the Internet is connected.

By comparing graphs for an individual site, one can get an impression whether one's side has a better (or worse) connectivity to the global Internet. One has to be a bit careful here. The distribution also depends on the global distribution of the Test Boxes, and sites in areas with only a few boxes (Asia, South America) will always appear to be worsely connected than sites in areas with a high number of boxes (Europe, North America).

Why it is Measured

These graphs a by-product of the TTM measurements that have global interest.

What You Can Do with It

Data is by Test Box and for the entire sample. By comparing the shape of graph for an individual Test Box against the global distribution, one can get an impression of how well one's site is connected to the Internet. In general, a lower average and a lower spread are better.

Data over time

One can also look at the data over time, sharp jumps indicate that upstreams were added (or lost) and better (worse) routing was possible.

Plans

This is a first step towards an analysis of the development of the routing graph over time. We are currently working our way backwards through older data, when that is done, we will revisit this analysis. Watch this page for further results.