Re: [Idr] I-D Action: draft-ietf-idr-tunnel-encaps-00.txt

[Robert Raszuk <[e-mail]>]

Hi Acee,

Of course I meant to include reference to EPE not the BGP-LS extension to
it :)
​Thank you !​


​The correct one is this:​

https://tools.ietf.org/html/draft-filsfils-spring-segment-routing-central-epe-04

​To be precise the section I am referring to in this doc is section 5 where
using an encap SAFI would be section 5.6 when one would send next hops in
the NLRI with SR attribute or Encap Attribute carrying the egress
engineered labels, GRE keys etc....

Note that how the packet is sent over the network from ingress to egress
would be orthogonal. .

Best,
Robert.


On Wed, Sep 2, 2015 at 9:03 PM, Acee Lindem (acee) <[e-mail]> wrote:

> Hi Robert,
>
> On 9/2/15, 11:50 AM, "Idr on behalf of Robert Raszuk"
> <[e-mail] on behalf of [e-mail]> wrote:
>
> >Hi Eric,
> >
> >> Of course, nothing stops someone from writing a new draft specifying
> >> something like the Encaps-SAFI, and specifying the semantics of
> >> attaching various attributes (including the tunnel encapsulation
> >> attribute) to it.
> >
> >True that nothing stops to do that work. It is only that such work to
> >go from draft to RFC will take easily 5-8 years in the current IETF
> >process.
> >
> >Use case for encapsulation SAFI I see is described in
> >https://tools.ietf.org/html/draft-ietf-idr-bgpls-segment-routing-epe-00
> >especially when there is need to do EPE in traditional WAN networks
> >and not those "trendy" ones Keyur referred to which no longer use ISIS
> >or OSPF.
>
> Are you sure you meant to reference the BGP EPE draft? AFAIK, there is no
> requirement for the encapsulation SAFI in this draft. The advertised EPE
> SIDs are local segment routing labels - MPLS is the implied encapsulation.
>
> Thanks,
> Acee
>
>
>
>
> >
> >As you know (apparently not everyone does) attaching an attribute to
> >route which is not used for forwarding (due to Admin Distance as
> >example) makes implementations very ugly and kludgy.
> >
> >Sorry but I do not buy arguments stating as a main reason that
> >deploying new SAFI is hard. If you deploy new SAFI in a p2p fashion
> >between controller and PE/ASBR (which is exactly what is needed for
> >EPE use case) its deployment is as easy as deploying BGP-LS SAFI.
> >
> >Cheers,
> >Robert.
> >
> >PS1:
> >
> >> But the feedback (both public and private) has been
> >> overwhelmingly in favor of obsoleting RFC5512
> >
> >Just curious if that feedback came from the same folks who voted and
> >encouraged original specification or completely different ones .... ;)
> >
> >PS2:
> >
> >For EPE use case using Encap SAFI for signalling controller to edges
> >is just an option. There could be of course alternatives worked for
> >example in I2RS WG or using YANG models and NETCONF. So I will not be
> >making much noise about keeping it any more. It just looks bizarre to
> >obsolete something which seems useful.
> >
> >_______________________________________________
> >Idr mailing list
> >[e-mail]
> >https://www.ietf.org/mailman/listinfo/idr
>
>