From: "Dmitri Krioukov" <dima@caida.org>
Date: 12 juli 2006 17.40.25 EDT
To: "'Avri Doria'" <avri@acm.org>
--
As pointed out in the charter, performance of any routing algorithm
strongly depends on the topological structure of an underlying
network:
almost all topologies are much better, with respect to routing,
than the
worst-case topology, while realistic, e.g., observed in the real
Internet,
topologies appear to be optimal (http://www.caida.org/~dima/pub/
crig-infocom.pdf).
This observation is the main reason why routing research in the last
two years has essential shifted its focus to topology analysis, in the
form of construction of better-scaling routing algorithms designed
specifically for narrower classes of more realistic topologies (see
the overview http://www.caida.org/~dima/pub/routing_perspectives.pdf)
Therefore, the focus of the RR-FS WG has remained on topology
analysis, with the main recent result to be presented at the next
SIGCOMM
http://www.caida.org/~dima/pub/dk.pdf Among other things, this work
reveals intricate and apparently profound relationships between local
and global topological characteristics of complex networks. This
finding
can potentially shed some light on how having only local topological
information, one can efficiently route through a global complex
network.
Other results include completing work on AS relationships,
http://www.caida.org/~dima/pub/as_rel.pdf
http://www.caida.org/data/active/as-relationships/
taxonomy,
http://www.caida.org/~dima/pub/as_char_pam.pdf
http://www.caida.org/data/active/as_taxonomy/
and comparative analysis of different sources of the Internet AS-level
topology data
http://www.caida.org/~dima/pub/as-topo-ccr.pdf
http://www.caida.org/analysis/topology/as_topo_comparisons/