Hi Mikael, > There are vendors that don't by default do this and I doubt they'll
Indeed ! And I think today it may actually not be that easy to accurately construct a filtering policy for EBGP peers. So with that IDR & GROW WGs (cc-ed) would fill about publishing on IDR or GROW wiki a csv file which would contain all attributes and communities which should be by default filtered on EBGP boundary (unless otherwise overwritten by operator for reasons like same admin domain as example)? It may be way easier than trying to analyze all drafts, temporary allocations and published rfc every day. It could be used as: * always up to date hint to the operator * direct import to NOS * input to AI to construct your BGP policy The format of such a csv file could be documented in a draft/rfc. It could be updated each time there is some new IANA allocation made in any of the BGP registry and adjusted when allocation changes. Alternatively it could be done outside of IETF (RIPE, NANOG etc ...) but then maintenance of it may be a bit harder. I think it also fits Jeff's attribute escape umbrella as one of the possible tools in the BGP toolkit. Thx, Robert On Sat, May 23, 2026 at 6:42 AM Mikael Abrahamsson <swmike= [email protected]> wrote: > On Fri, 22 May 2026, Keyur Patel wrote: > > > * In both cases, this filtering MUST be enabled by default for EBGP > sessions. > > There are vendors that don't by default do this and I doubt they'll > change, so then the next-best I can see is clear operational > recommendations to operators on what to do at their border, and at least > that the vendors have an easy way to achieve this. > > Right now I have one vendor that doesn't seem to make it easy to "filter > all extended communities on border". > > -- > Mikael Abrahamsson email: [email protected] > > _______________________________________________ > Idr mailing list -- [email protected] > To unsubscribe send an email to [email protected] >
_______________________________________________ GROW mailing list -- [email protected] To unsubscribe send an email to [email protected]
