[pim] Re: I-D Action: draft-ietf-pim-multicast-lessons-lea rned-06.txt

Let me answer Dave's comment another way:

(1) Hot potato routing in unicast means a unicast packet will exit an ISP as soon as possible. That is their unicast forwarding policy.

(2) Having said (1), the path is based on the unicast RIB. So a PIM Join message follows that hot potato path from recievers to the source.

Therefore, if an ISP chooses to have a congruent (RPF) topology, then the multicast will have similar paths. But having said that, the replication factor should be more top of mind. 

From my experience deploying multicast at cisco, most ISPs did not want to be "packet duplicators" so the branch of the distribution tree was a single "one path in one path out" and replication was pushed towards the edges (or more to the point, closer to receivers).

Dino

> On Nov 5, 2025, at 7:16 AM, Toerless Eckert <[email protected]> wrote:
> 
> I don't think it's that trivial. 
> 
> In an N-AS interdomain path AS1...ASN, if every AS plays hot potato,
> the unicast packet could only cool off in ASN based on ASN topology - because
> all prior AS kept it as hot as they could.
> 
> If Unicast AS1->ASN arrive hot, then multicast ASN->AS1
> arrives hot. If Unicast AS1->ASN arrives cold, then multicast
> ASN->AS1 arrives cold.
> 
> "If you're an AS that delivers cold potato unicast,
> then you are also an AS that sends chold potato multicast."
> 
> Of course, looking at my qualification, i only rank middle of the
> road for per capita potato consumption by passports. We'd need an eastern european
> or dutch person to chime in here. Although i think the dutch get
> up that high only with chips, so they may need to be disqualified for
> hot potato expertise.
> 
> Cheers
>   Toerless
> 
> On Sat, Oct 18, 2025 at 09:08:46AM -0400, David R. Oran wrote:
>> Thanks - this is a really nice contribution - represents a lot of work and perspective.
>> 
>> There’s one factor for the multi-domain adoption of multicast that I think was pretty important but wasn’t mentioned:
>> 
>> Aside from traffic engineering optimizations, ISPs as much as possible do “hot-potato” routing which tries to get the packet off their network as soon as possible. In contrast, multicast turns that on its head and causes “cold-potato” routing where the packet stays on the ISPs network as far as possible before exiting, thus sucking resources.
>> 
>> DaveO.
>> 
>> On 17 Oct 2025, at 21:08, [email protected] wrote:
>> 
>>> Internet-Draft draft-ietf-pim-multicast-lessons-learned-06.txt is now
>>> available. It is a work item of the Protocols for IP Multicast (PIM) WG of the
>>> IETF.
>>> 
>>>   Title:   Multicast Lessons Learned from Decades of Deployment Experience
>>>   Authors: Dino Farinacci
>>>            Lenny Giuliano
>>>            Mike McBride
>>>            Nils Warnke
>>>   Name:    draft-ietf-pim-multicast-lessons-learned-06.txt
>>>   Pages:   23
>>>   Dates:   2025-10-17
>>> 
>>> Abstract:
>>> 
>>>   This document gives a historical perspective about the design and
>>>   deployment of multicast routing protocols.  The document describes
>>>   the technical challenges discovered from building these protocols.
>>>   Even though multicast has enjoyed success of deployment in special
>>>   use-cases, this draft discusses what were, and are, the obstacles for
>>>   mass deployment across the Internet.  Individuals who are working on
>>>   new multicast related protocols will benefit by knowing why certain
>>>   older protocols are no longer in use today.
>>> 
>>> The IETF datatracker status page for this Internet-Draft is:
>>> https://datatracker.ietf.org/doc/draft-ietf-pim-multicast-lessons-learned/
>>> 
>>> There is also an HTMLized version available at:
>>> https://datatracker.ietf.org/doc/html/draft-ietf-pim-multicast-lessons-learned-06
>>> 
>>> A diff from the previous version is available at:
>>> https://author-tools.ietf.org/iddiff?url2=draft-ietf-pim-multicast-lessons-learned-06
>>> 
>>> Internet-Drafts are also available by rsync at:
>>> rsync.ietf.org::internet-drafts
>>> 
>>> 
>>> _______________________________________________
>>> I-D-Announce mailing list -- [email protected]
>>> To unsubscribe send an email to [email protected]
>> 
>> DaveO
> 
> _______________________________________________
> pim mailing list -- [email protected]
> To unsubscribe send an email to [email protected]

_______________________________________________
pim mailing list -- [email protected]
To unsubscribe send an email to [email protected]