Re: [Lsvr] Comments on LSVR with RR peering

[Randy Bush <[e-mail]>]

>>> Perhaps the authors can clarify what they meant by "BGP speakers peer
>>> solely with one or more ... controllers".
>>> 
>>> On 12/5/2018 3:58 PM, Randy Bush wrote:
>>>> i think i see the disconnect
>>>> 
>>>>>>> RR--1--A--1--B--1--C--1--D--1--E
>>>>>>>                  |                 |
>>>>>>>                  |-----5-----F--1--|
>>>>>>> 
>>>>>>> The letters represent routers (RR being the Route Reflector), and the
>>>>>>> numbers are link metrics.
>>>> B, C, D, E, and F may not be clients of RR.  2.2 says you can use
>>>> loopbacks.  this does not imply that you can multi-hop peer a la ibgp.
>>>> hence, 2.3 does not allow multi-hop.
>>>> 
>>>> or i am still confused.
>>>> 
>>>> randy
>> perhaps they thought "directly" in the section title of 2.2 would give
>> us a clue
>> 
>>       2.2. BGP Peering Between Directly Connected Network Nodes
>> 
>> followed by
>>         2.3.  BGP Peering in Route-Reflector or Controller Topology
>> 
>>         In this model, BGP speakers peer solely with one or more Route
>>         Reflectors [RFC4456] or controllers.  As in the previous model,
>>         direct connection discovery and liveliness detection for those
>>         connections are done outside the BGP protocol.
>> 
>> so would you have them clarify 2.3 with something such as
>> 
>>      as with non-spf bgp route reflection, clients are directly connected
>>      to the reflector(s)
>> 
>> randy
> 
> Presumably in needs to be "directly connected via an alternate
> physical network."

alternate to what, exactly?

do not be fooled by the diagram above, which is not a route reflector
cluster.

randy