On Tue, Nov 17, 2020 at 9:43 AM Robert Raszuk <rob...@raszuk.net> wrote:
> Hi Gyan,
>
> Gyan>. We could use Aijun’s passive interface new top level TLV to
>> link the next hop rewrite loopback to the PE-CE links all being set to
>> passive. So if any PE-CE link goes down a PUA is sent and the next hop
>> converges PIC core PE-CE link which is now associated with the Loopback.
>> This would be a major benefit of PUA for PIC core convergence when
>> next-hop-self is used which applies to MPLS and SR and IP based core.
>>
>
> I have read the above three sentences five times and came with two basic
> observations:
>
> * You are mixing PIC Core with PIC Edge - Please do not. Those are
> completely separate features and it is a feature not a bug to keep them
> that way.
>
Gyan> Sorry for the confusion. I agree we are not taking BGP PIC core
which is H-FIB core feature for core failures resulting in IGP next hop
update. Also we are not talking about BGP PIC edge pre programmed backup
paths. What I was trying to address is use of PUA to resolve a issue with
BGP convergence related to using BCP use of next hop self rewrite so PE-CE
link next hops do not have to be flooded into the IGP. Down side of
next-hop-self is for convergence in the core from edge failure, the edge
failure is not detected and you have to wait for route to be withdrawn. So
I was thinking maybe PUA data plane convergence mechanism would be a way to
help with convergence. Not sure how or if possible but was thinking that
as a possible use case for PUA.
>
> * You are mixing external PE-CE interfaces with IGP passive interfaces -
> Please do not. There should be no IGP flooding of any sort associated with
> state of PE-CE interface. No need.
>
Understood. Bad idea.
>
> Thx a lot,
> R.
>
> --
<http://www.verizon.com/>
*Gyan Mishra*
*Network Solutions A**rchitect *
*M 301 502-134713101 Columbia Pike *Silver Spring, MD
_______________________________________________
Lsr mailing list
Lsr@ietf.org
https://www.ietf.org/mailman/listinfo/lsr
