Hi,

> Considering what RFC8212 does for prefixes, I'd like to discuss doing the
> same for communities (incl extended and large).

I think that could be easily done.

But what about BGP Attributes which should never be sent interdomain ? As
you can see from csv there are lots of them and the list keeps growing.

I think if not already then soon we may also need clarity on which
AFI/SAFIs should be enabled in the true Interdomain case and which should
be rather avoided.

Thx,
R.

On Sat, May 23, 2026 at 5:31 PM Mikael Abrahamsson <[email protected]> wrote:

> On Sat, 23 May 2026, Robert Raszuk wrote:
>
> > It is highly amazing how much "stuff" carried by BGP today is not
> > recommended to be used inter-domain !!!
>
> Considering what RFC8212 does for prefixes, I'd like to discuss doing the
> same for communities (incl extended and large). If a policy doesn't
> modify/touch/mention extended/large communities, then default should be to
> drop these on accepted prefixes.
>
> Controlling this behavior could be either a global setting, a peer-group
> setting, or a per-neighbor setting.
>
> Also, considering where a lot of the datacenter is going, differentiating
> on EBGP/IBGP is not as effective today as it used to be, so forcing the
> user to set up a policy to propagate routes (RFC8212), but also have that
> policy decide (default drop/permit) what to do about communities, would be
> a good thing.
>
> I've seen DC implementors who definitely want "full open, no filtering" on
> EBGP, but for the ISP inter-network space, allowing extended/large by
> default is dangerous.
>
> --
> Mikael Abrahamsson    email: [email protected]
>
_______________________________________________
GROW mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to