Re: [pim] [Bier] [bess] WGLC: draft-ietf-bier-mvpn-05

[Eric C Rosen <[e-mail]>]

On reflection, I think it's best if there is no requirement for a PE's 
BFR-Prefix to appear in the NLRI of any of the MVPN routes, no 
requirement for it to appear in the VRF Route Import EC, no requirement 
for it to appear in any next hop field, no requirement for it to appear 
in the Inter-Area P2MP Next Hop EC, etc.  That is, no requirement for it 
to be anything other than a BFR-Prefix.

So if the PTA of an x-PMSI A-D route specifies tunnel type "BIER", I 
think the simplest thing is to require that the BFR-Prefix be specified 
in the PTA.

Similarly, the simplest way to ensure that the ingress PE knows the 
BFR-Prefix of an egress PE is to have the egress PE include a PTA in the 
Leaf A-D route, and require to PTA to specify "tunnel type" BIER, MPLS 
Label 0 (ignored), and the BFR-Prefix of the Leaf.  Which I think was 
Stig's original suggestion.


On 6/21/2017 1:08 PM, Mahesh Sivakumar (masivaku) wrote:
> Hi Eric,
>
>>> For BIER, I was thinking that the BFR-Prefix of the egress PE should
>>> appear in the "originating router's IP address" field of the Leaf A-D
>>> NLRI.  However, it is probably better to allow the "originating router's
>>> IP address" to be different than the BFR-Prefix, and in that case to use
>>> the Leaf A-D route's PTA to specify the BFR-Prefix.
> I was thinking along the same lines so that we have the flexibility on the
> Ingress BFR to use either the originating router or the BFR prefix while
> tracking.
> But do you envision a real use case at the ingress BFIR for using the BFR prefix
> in the PTA given that the BFIR can use the originating router IP address to
> match the appropriate S-PMSI A-D route.
> Accordingly, should we make the BIER-pfx in the PTA optional?
>
> Thanks
> Mahesh
>
>
>
>
>
> On 6/21/17, 8:39 AM, "BIER on behalf of Eric C Rosen" <[e-mail] on behalf of [e-mail]> wrote:
>
>> Stig, thanks for your comments.+
>>
>> On 6/19/2017 5:47 PM, Stig Venaas wrote:
>>> Hi
>>>
>>> I think this draft is mostly ready. I just have a couple of comments.
>>>
>>> In section 1:
>>>      This revision of the document does not specify the procedures
>>>      necessary to support MVPN customers that are using BIDIR-PIM.  Those
>>>      procedures will be added in a future revision.
>>>
>>> Remove this text?
>> We'll probably just change this to something like "Procedures to support
>> MVPN customers that are using BIDIR-PIM are outside the scope of this
>> document".
>>
>>> Section 2.1.  MPLS Label
>>>
>>> Should one use different labels to distinguish address families in the same VRF?
>> Nice catch.  The customer's address family is identified by the AFI of
>> the MCAST-VPN routes.  There should be a requirement that a given router
>> MUST NOT originate two x-PMSI A-D routes with different AFIs but with
>> the same upstream-assigned label in their respective PTAs.
>>
>>> The PTA must be present in Leaf A-D routes so one can know the BIER
>>> prefix of the router joining. It might be obvious, but I think it is
>>> worth pointing it out. It is specified for IR (in RFC 7988 section
>>> 4.1.1 it says: "Leaf A-D route MUST also contain a PTA"...
>> For IR, the PTA is needed because each egress PE needs to advertise a
>> downstream-assigned label.
>>
>> For BIER, I was thinking that the BFR-Prefix of the egress PE should
>> appear in the "originating router's IP address" field of the Leaf A-D
>> NLRI.  However, it is probably better to allow the "originating router's
>> IP address" to be different than the BFR-Prefix, and in that case to use
>> the Leaf A-D route's PTA to specify the BFR-Prefix.
>
>
>> _______________________________________________
>> BIER mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/bier