draft-ietf-mpls-sfl-control-03 (AD Evaluation:: Revised I-D Needed)
", "time": "2023-11-09T12:02:41Z"}, {"author": "Stewart Bryant", "text": "Revised ID uploaded last Monday
", "time": "2023-11-09T12:03:04Z"}, {"author": "Dhruv Dhody", "text": "Codimd link on the slide \u2014 blast from the past :smile:
", "time": "2023-11-09T12:03:12Z"}, {"author": "Adrian Farrel", "text": "Eeek. Good point.
", "time": "2023-11-09T12:03:48Z"}, {"author": "Stewart Bryant", "text": "+1 - thanks Loa
", "time": "2023-11-09T12:05:30Z"}, {"author": "Andy Malis", "text": "+1 as well
", "time": "2023-11-09T12:05:44Z"}, {"author": "Ahmed Bashandy", "text": "+1 for Loa
", "time": "2023-11-09T12:05:47Z"}, {"author": "Lou Berger", "text": "Loa - thank you for your contributions and leadership
", "time": "2023-11-09T12:05:59Z"}, {"author": "Lou Berger", "text": "MPLS wouldn't be what it is without you
", "time": "2023-11-09T12:06:25Z"}, {"author": "Stewart Bryant", "text": "Publication has the advantage of wider review of the work
", "time": "2023-11-09T12:33:17Z"}, {"author": "J\u00e1nos Farkas", "text": "If there is/are questionable use cases, then may be better to remove it/them before publication.
", "time": "2023-11-09T12:37:09Z"}, {"author": "Stewart Bryant", "text": "sometimes use cases are ahead of th established method
", "time": "2023-11-09T12:38:18Z"}, {"author": "Stewart Bryant", "text": "maybe we remove the detent reference but keep the use case
", "time": "2023-11-09T12:38:47Z"}, {"author": "Lou Berger", "text": "the most that could be said at this point is a personal opinion
", "time": "2023-11-09T12:40:00Z"}, {"author": "Lou Berger", "text": "IMO
", "time": "2023-11-09T12:40:17Z"}, {"author": "Tony Li", "text": "MNA is extensible...
", "time": "2023-11-09T12:40:33Z"}, {"author": "Stewart Bryant", "text": "I am sure I can find an ITU document that calls up the latency use case
", "time": "2023-11-09T12:42:03Z"}, {"author": "Eduard V", "text": "I was surprised too that the list of use cases is so short. SPRING is much better on marketing side. You would not sell this elephant with such marketing.
", "time": "2023-11-09T12:42:03Z"}, {"author": "Adrian Farrel", "text": "It's an elephant? Not a camel?
", "time": "2023-11-09T12:45:01Z"}, {"author": "Tony Li", "text": "I was wrong, it's a rock. Camels and elephants move.
", "time": "2023-11-09T12:45:32Z"}, {"author": "J\u00e1nos Farkas", "text": "Sorry if I confused the terms \"use case\" and \"solution\". Having timing information in packet header is a solution from DetNet's perspective. DetNet WG has not reached consensus on what solution to go for to meet scaling requirements.
", "time": "2023-11-09T12:46:27Z"}, {"author": "Eduard V", "text": "I know proverb about difficulty to sell elephant not camel or rock.
", "time": "2023-11-09T12:47:20Z"}, {"author": "Jie Dong", "text": "Just took a quick look at the use case draft, some use cases (e.g. GDF) are not listed in the slides, are they still considered valid?
", "time": "2023-11-09T12:48:39Z"}, {"author": "Tony Li", "text": "IMHO, the only thing blocking MNA right now is a call about whether PSD is required right now. The chairs should start that consensus call now.
", "time": "2023-11-09T12:49:37Z"}, {"author": "Tarek Saad", "text": "@Jie - authors recently dicussed with GDF proposal and the conclusion is GDF is no longer pursued. So, this is one use case that will be removed as the last slide mentioned
", "time": "2023-11-09T12:50:01Z"}, {"author": "Tony Li", "text": "If PSD is not required right now, then it can be deferred until when we have a compelling use case.
", "time": "2023-11-09T12:50:07Z"}, {"author": "Tony Li", "text": "The docs then may need some wordsmithing and are otherwise ready to go.
", "time": "2023-11-09T12:50:34Z"}, {"author": "Stewart Bryant", "text": "Agree - next step is to determine if PSD is in the current design or postponed
", "time": "2023-11-09T12:52:09Z"}, {"author": "Adrian Farrel", "text": "@tony. At the very least, the requirements draft needs an update for the many comments received on list. So everyone can be busy polishing documents to make the proress smoother
", "time": "2023-11-09T12:52:32Z"}, {"author": "Loa Andersson", "text": "when we discussed \"use cases\" we said that requirements are \"derived\" from the use cases. A use case that does not generate new reqirements are not high priority. I agree with Tony that is shipable today. The intresting question is whether the intended \"harbour\" is willing to receive it.
", "time": "2023-11-09T12:52:57Z"}, {"author": "Stewart Bryant", "text": "One suggesting is that we split the REQ and FWK into two publish ISD now and publish PSD when we are ready
", "time": "2023-11-09T12:53:55Z"}, {"author": "Loa Andersson", "text": "Maybe we should call the PSD question, that horse is beaten to death
", "time": "2023-11-09T12:54:09Z"}, {"author": "Robert Raszuk", "text": "Sorry have not been following MNA story too closely recently, but is the plan to still use (overload) the same MPLS ethertype(s) ? If so why not allocate new one and leave current MPLS as is ?
", "time": "2023-11-09T12:56:18Z"}, {"author": "Tony Li", "text": "Yes, same ethertype.
", "time": "2023-11-09T12:56:47Z"}, {"author": "J\u00e1nos Farkas", "text": "sorry
", "time": "2023-11-09T12:57:13Z"}, {"author": "Tony Li", "text": "A new ethertype would take 5 years to get deployed.
", "time": "2023-11-09T12:57:16Z"}, {"author": "J\u00e1nos Farkas", "text": "audio does not work
", "time": "2023-11-09T12:57:17Z"}, {"author": "Stewart Bryant", "text": "Agree with Tony
", "time": "2023-11-09T12:57:47Z"}, {"author": "Tony Li", "text": "@Robert Raszuk There is no obligation to use MNA. If you don't like it, don't use it.
", "time": "2023-11-09T12:58:05Z"}, {"author": "Robert Raszuk", "text": "Not sure why 5 years .. you still need bits to recognize new protocol elements .. they were just sit closer to the wire.
", "time": "2023-11-09T12:58:14Z"}, {"author": "J\u00e1nos Farkas", "text": "I'd repeat what we said: DetNet WG has a WG document on scaling requirements, but no other WG document yet
", "time": "2023-11-09T12:58:21Z"}, {"author": "J\u00e1nos Farkas", "text": "no solution developed
", "time": "2023-11-09T12:58:29Z"}, {"author": "J\u00e1nos Farkas", "text": "we have meetings to work on solution to address scaling requirements
", "time": "2023-11-09T12:58:48Z"}, {"author": "Stewart Bryant", "text": ".. and it would be a silly to need to do a forklift network upgrade for end to end MNA
", "time": "2023-11-09T12:58:54Z"}, {"author": "Robert Raszuk", "text": "Encap usually helps to avoid end to end network upgrades when we roll any new protocol
", "time": "2023-11-09T12:59:50Z"}, {"author": "J\u00e1nos Farkas", "text": "I agree with what Lou says.
", "time": "2023-11-09T12:59:57Z"}, {"author": "Louis Chan", "text": "Any change in forwarding plane is an expensive exercise from $ point of view
", "time": "2023-11-09T13:00:34Z"}, {"author": "Tony Li", "text": "I don't understand the issue. Detnet uses MNA when it wants. No rush...
", "time": "2023-11-09T13:00:34Z"}, {"author": "Tony Li", "text": "@Robert Raszuk A new encap would require an end-to-end upgrade to use MNA.
", "time": "2023-11-09T13:02:00Z"}, {"author": "Stewart Bryant", "text": "exactly
", "time": "2023-11-09T13:02:15Z"}, {"author": "Robert Raszuk", "text": "not if the encap is MPLS 8847/8848
", "time": "2023-11-09T13:03:48Z"}, {"author": "Loa Andersson", "text": "wouldn't it be the case that not all packets in a flow capackets belonging to the same flow be on different Ethertypes depending on MNA or not?
", "time": "2023-11-09T13:06:34Z"}, {"author": "Stewart Bryant", "text": "Yes that is a point not all packets in a flow will have MNA
", "time": "2023-11-09T13:07:30Z"}, {"author": "Robert Raszuk", "text": "I guess it depends on the application. For some SR analogy it would work fine .. for others counting on new actions on every node - every node needs to be upgraded anyway
", "time": "2023-11-09T13:08:00Z"}, {"author": "Robert Raszuk", "text": "irrespective if only 1% of packets in the flow need MNA
", "time": "2023-11-09T13:08:38Z"}, {"author": "Jie Dong", "text": "Although the detnet use case and solution may not be stable at current stage, it shows that for some cases ISD has some limitations, and PSD could be more efficient
", "time": "2023-11-09T13:10:05Z"}, {"author": "Tony Li", "text": "? I didn't see that at all.
", "time": "2023-11-09T13:11:11Z"}, {"author": "Jie Dong", "text": "It is quoted from the detnet slides just presented
", "time": "2023-11-09T13:11:43Z"}, {"author": "Louis Chan", "text": "@Robert, agree. unless there is a upcoming scaling issue that cannot be solved today with 20bit label.
", "time": "2023-11-09T13:12:39Z"}, {"author": "Tony Li", "text": "@Jie Dong I'm not seeing that in the slides. Could you please be more specific?
", "time": "2023-11-09T13:13:33Z"}, {"author": "Robert Raszuk", "text": "My major point is that popularity of MPLS is due to two facts: simplicity, at the time of deployment no alternative line rate encap. Extensibility was never big on the MPLS agenda and extensibility (especially when we talk data plane) is enemy of simplicity.
", "time": "2023-11-09T13:15:38Z"}, {"author": "Tony Li", "text": "And here we are, 25 years later, finding that we need extensibility. This seems like a fine lesson.
", "time": "2023-11-09T13:16:45Z"}, {"author": "Joel Halpern", "text": "Even if some use case, if needed, might be more fficient in PSD encoding, that doesn't drive doing PSD now. It would be an argument for PSD when and if detnet says they need. Let's get ISD done now to meet the needs we have.
", "time": "2023-11-09T13:17:01Z"}, {"author": "Jie Dong", "text": "@tony, it is on page 5 of the detnet slides
", "time": "2023-11-09T13:17:41Z"}, {"author": "Tony Li", "text": "Do you understand the word 'MAY'?
", "time": "2023-11-09T13:18:03Z"}]