Re: [sidr] Injecting idea of "freshness of repository data" into BGP

[Terry Manderson <[e-mail]>]

Jeff,

On 28/03/12 6:19 PM, "Jeffrey Haas" <[e-mail]> wrote:

> Per my mic comment at IETF 83:
> During the San Diego interim session we had discussed potentially signaling
> in BGP the idea that a given AS may have fresher data available in its
> repository.
> 
> My original thought had been something along the lines of a new AFI/SAFI
> that contains this data.  Matt L., in discussing this point at the mic with
> me, suggested something that has resemblence to the serial number field in
> DNS.  For example, this field could go into the "reserved" field that a
> route originator puts into the signature.  If the serial number increases,
> this could suggest that fresher information is present in that originator's
> repository.

I think this is interesting. I think I would further like an
assessment/disussion of this "serial number" being consistent between the
BGP information, the RPKI repository, and this through the validated cache
and presented to the router via rpki-rtr.

It may well present far too many error situations by doing that, but may
also provide a brilliant statement of a consistent view matching origination
intent in a time and space perspective.

0.02c

> 
> Discussion around this mechanism:
> - If this is part of a given route's signature block, the issue is that a
>   given origin may be seen with a number of serial numbers  depending on
>   propagation of BGP routes.
> - Such a serial number is important not only for the originator of a route,
>   but also all parties participating in the signature.
>   This particular details suggests to me that such signaling probably should
>   be separate from the signatures.
> - By being part of the signature, we get some level of freshness in things
>   in a route-by-route basis and less likely that a completely separate
>   "route" that is repository freshness is dropped.
>