Hi Taral, > Message-ID: <fa0147d90905291236v38e56e11h12cb2033f7cd7...@mail.gmail.com> > From: Taral <tar...@gmail.com> > Date: Fri, 29 May 2009 12:36:26 -0700
> I remember years ago being thoroughly impressed with Koorde's > theoretical efficiency. Does anyone know of a system that implemented > it? Did they achieve anything close to the described efficiency? Overlay Weaver includes its implementation. Overlay Weaver: An Overlay Construction Toolkit http://overlayweaver.sf.net/ I recently measured the path lengths of routing results. Routing path length http://www.shudo.net/diary/data/20090506/path-length.png The result shows relatively long path length of Koorde instead of small routing table. I noticed that Koorde's routing works well in a stable situation but it has difficulty with churn and incomplete routing tables. In Koorde's de Bruijn routing, the next hop is not necessary to be close to the target ID in the mean of ID distance (clock-wise difference between IDs). Once a node fails to find an living next hop in its de Bruijn edges and adopts a successor as the nexthop, de Bruijn routing so far becomes completely useless. The routing uses successor chain to the target ID or starts de Bruijn routing from beginning again. Kazuyuki Shudo 2...@shudo.net http://www.shudo.net/ _______________________________________________ p2p-hackers mailing list p2p-hackers@lists.zooko.com http://lists.zooko.com/mailman/listinfo/p2p-hackers
