AS-set results, new experiments 30/06/2005


we have posted preliminary results of our long AS-set announcements to
our web site:

While we are still working on an in-depth analysis, the results indicate that while the presence or absence of an AS-set in an announcement does have an effect on its propagation, the length of the AS-set, if present, does not have a significant effect.

We would also like to move forward with further announcements to determine the effectiveness of our topology discovery algorithms in the IPv4 network. The algorithms, described in our posting a few weeks ago [1] and on the project's web page [2], make use of announcements involving the AS numbers of other ASes to allow ISPs to discover how their announcements can be propagated and how the policies of other ASes in the Internet affect their prefixes.

The extensive tests we have performed on the global IPv6 Internet show that our algorithms are effective, but we would like to see how effective they are on the IPv4 Internet given the different operational practices in use. Therefore, starting on Thursday 30 June at 14:00 UTC the prefixes and will be announced with AS-paths of the form 12654 {...} where the AS-set contains the numbers of neighboring ASes.

The announcements will not last more than a few days; as before, should anyone see any problems, please contact me and I will take immediate action. Finally, our tests show that our announcements cause no technical problems; however, if anyone would prefer their AS number not to appear in the AS-sets, and feels that filtering our prefixes is not sufficient, please contact me and we will not use them.