Re: [sidr] 4-byte vs 2 byte ASN (was re: I-D Action: draft-ietf-sidr-pfx-validate-03.txt)

[John Leslie <[e-mail]>]

George, Wes <[e-mail]> wrote:
> 
> I'm thinking that we need a comment early in the draft stating that
> for the remainder of the draft no distinction is being made between
> AS_PATH and AS4_PATH,

   I wouldn't phrase it that way -- I'd emphasize instead that the AS
Path being validated is the AS Path _after_ reconstruction using
AS4_PATH in the case where routing information comes from a peer which
doesn't recognize 4-byte ASNs.

   (It would be simply wrong to "validate" AS4_PATH in the case where
you don't also do the reconstruction, or in the protocol-violating
case where you receive AS4_PATH from a peer which does speak 4-byte
ASNs.)


> Or alternatively, specify that this validation is performed on
> AS4_PATH

   That would also be wrong. :^(

> and require support for 4893 as a prerequisite for SIDR.

   IMHO, support for 4-byte ASNs should be a prerequisite. Alas, this
_does_ need to be stated.

> If we don't explicitly require hosts that support SIDR origin
> validation to support 4-byte ASN, we may also need some direction
> regarding specific handling for AS23456,

   It would be good to add that anyway, since I contend that we can't
be sure AS23456 will always be removed.

--
John Leslie <[e-mail]>