You can find an analysis of the utilization of communities found in routing tables collected by RIPE RIS and RouteViews at http://www.infonet.fundp.ac.be/doc/reports/Infonet-TR-2002-02.pdf. In this analysis we show two things: (1) communities tend to be widely used and (2) communities are used for route tagging (for instance to remember where a route has been issued and traffic engineering purposes (for instance to influence how a peer will redistribute our routes).
The results of the analysis are available from http://alpha.infonet.fundp.ac.be/anabgp By the way, we have presented our work during the last NANOG meeting in Toronto. The slides are available from http://www.nanog.org/mtg-0206/bruno.html Bruno. Thomas Kernen wrote: > >This started off as me being curious as to why a UUNet engineer I was >talking to told me he could not understand why a network would support a >feature such as BGP communities for identifying the origin of a >route/network entry point. I tried to explain to him the advantage of being >able to quickly identify where a route originates from (geographically), >type of interconnect, type of "peer" (in this case I use peer for any BGP >peer, customer or transit). I explained that it could be usefull for >debugging and gaining more background info (route analysis is one of my >favorite tasks) and some of the major and minor networks do provide such a >feature/service. > >Still the engineer could not understand why and only saw this as a security >issue, well I guess when you work for a network that does not provide any >public looking glass or route server it's not really a surprise </rant> > >This triggered a thought, do many people actually use BGP communities to >pinpoint a route origination point/type, and if so for what purpose >(debugging, analysis, other) > >Thomas > >PS: If UUNet do actually support this feature please tell me who I should >contact. >
