IETF 108 Spring WG Meeting Minutes

Monday, 11:00-12:40, July 27, 2020 (UTC)

o SPRING Status [ 15 minutes ]
- Note Well
- Scribe
- Document Status

During the chair presentation, the SR over IPv6 Compression Design Team deliverables were discussed. It was clarified that the chairs are not asking the DT to select a solution, but after producing requirements to compare the solutions.

Cheng Li (Huawei): The second draft is an analysis draft instead of a comparison draft. Confirmed by Bruno.

o Segment Routing Policy Architecture [ 10 minutes ]
Ketan Talaulikar

Bruno: It is important to review this draft since it is an important draft for the WG.

o Service Programming with Segment Routing [ 5 minutes ]
Francois Clad

-no comments-

o SR over IPv6 compression - Design Team work plan [ 5 minutes ]
Weiqiang & Sander

Bruno: It would be better to have the first result presented by 15th September.
Andrew Alston: Comment on the time frame. Nine-month is a long time,but operators are desperate about the outcomes, whether there is way to shorten the time consumed.
Weichang: The time frame has been agreed by the team. To work more efficiently may help.

o Export of MPLS Segment Routing Label Type Information in [ 5 minutes ]
IP Flow Information Export (IPFIX)
Thomas Graf

Bruno: A short email to the lsr WG regarding the IANA allocation for ospfv2 and ospfv3 whether one for each or just one.

o Seamless Segment Routing [ 10 minutes ]
Shraddha Hegde

Jie Dong: SR is used as one type of the tunnels, RSVP-TE can also be used, why is it called as seamless SR?
Shraddha: Interworking with other SR technologies, this is the reason for the name.
Jie: But it can be more broader than SR.
Shraddha: We can think about renaming it.
Jie: There are extensions in the IDR WG. Is this draft a STDs or Info?
Shraddha: it goes to STDs track, including architecture and requirements.
Jie: Any changes to the SR architecture?
Shraddha: No. It is mainly about how to connect multiple domains with BGP-based mechanisms.

o Point-to-Multipoint Transport Using Chain Replication in Segment Routing [ 10 minutes ]
Yimin Shen

-no show and no time left at the end-

o Enhanced Performance Delay and Liveness Monitoring in Segment Routing Networks [ 10 minutes ]
Rakesh Gandhi

Greg Mirsky: What are you Monitoring?
Rakesh: SR Path. Probes follow the same data path of the SR Policy segment-list.
Greg: Is it to replace BFD?
Rakesh: It is enhanced monitoring, using the existing TWAMP mechanisms at the same time.
Greg: How do you check continuity?
Rakesh: Probe message along the SR path. If lost messages, liveness declared failed. Simple TWAMP leveraging, the packet format can be the same or different.
Greg: confusing. How Liveness scope is defined? Neither TWAMP or STAMP has the information.
Rakesh: Reflector is unaware of TWAMP or STAMP in loopback mode. Just using its message format.
Greg: how the reflector know the rate
Rakesh: There is no signalling, all based on the local provisioning. Probe packets are fast forwarded, no state or knowledge of the protocol on reflector. Rate limiting is not needed as packets are not punted on the reflector.
Greg: path continuity, the session is up. confusing statement.
Rakesh: Probes simply fast forwarded by the reflector that is why it is unaware of the state.
Greg: loopback mode is confusing.
Rakesh: Reflector just forwards packets in loopback mode without punting and re-injecting.
Greg: TWAMP and STAMP do not have liveness.
Rakesh: It does not have to be based on the existing mechansims
Greg: new protocols?
Rakesh: reusing TWAMP packet format. It’s not forbidden!
Bruno: please continue on the mailing list.

Andrew: can see a lot of use cases, especially actual timing across the path. It is currently tied to the TWAMP and STAMP at this moment in Section 5.2.
Rakesh: will consider updating the section. Using PM probes for delay, Liveness comes for free.

o Performance Measurement Using STAMP for Segment Routing Networks [ 10 minutes ]
Rakesh Gandhi

Greg: 1. introducing the control code, encourage the SPRING Chairs to discuss with the IPPM Chairs.
2. standalone LM message, why not use rfc 8762 and sequence number?
Rakesh: 1. have been keeping IPPM WG in the loop. Also, presenting the updates to IPPM on Friday.
2. meeting got disconnected.

o Performance Measurement Using TWAMP Light for Segment Routing Networks [ 10 minutes ]
Rakesh Gandhi

-no time to present-