IETF 114 RTGWG Meeting Minutes Chairs: Jeff Tantsura (jefftant.ietf@gmail.com) Yingzhen Qu (yingzhen.ietf@gmail.com) WG Page: http://tools.ietf.org/wg/rtgwg/ Materials: https://datatracker.ietf.org/meeting/114/session/rtgwg Meeting Administrivia and WG Update Chairs (5 mins) ============================================= WG document Update YANG Models for Quality of Service (QoS) Aseem Choudhary. (5 mins) https://datatracker.ietf.org/doc/draft-ietf-rtgwg-qos-model/ No questions at the end. Dynamic Networks to Hybrid Cloud DCs Problem Statement and Gap Analysis Linda Dunbar (10 mins) https://datatracker.ietf.org/doc/draft-ietf-rtgwg-net2cloud-problem-statement/ https://datatracker.ietf.org/doc/draft-ietf-rtgwg-net2cloud-gap-analysis/ [Alvaro] Speaking as a WG participant, I think documents like this are important as they will guide the solutions. It is very good to document the problems and gaps to guide work in various WGs in Routing area. However, if we publish the gaps as RFC, then they become frozen. When solutions become available, the gaps no longer exist. I’d suggest we maintain these documents as working documents in the WG and maintain periodically, so they remain useful. [Yingzhen] there is new information added, so please review it. [Linda] There are few use cases added. [Tianji] slides #4, there is a third case, the cloud provider may have its own backbone, in that case you may not have so many failures. RFC 5798bis Acee Lindem (10 mins) [Andrew Routing AD] Big thank you for this. Please keep going [Alvaro Routing AD] It is not requirement to publish the implementation status to move to standard. The main thing is to keep the changes minimal. make sure to document those changes in the shepherd review. [Acee] I will revise the draft. Many of the changes are editorial. I can keep it minimal. ============================================= Revived WG document: Destination/Source Routing Jen Linkova (10 mins) https://datatracker.ietf.org/doc/draft-llsyang-rtgwg-dst-src-routing/ [Adrian Farrel] No objection to the WG picking this up again. Thanks for long section on interop issues, and for calling out concerns about forwarding loops. I think this could be called “Semantic Routing” because you are making a forwarding decision based on something more than just the destination address - and that’s OK. But maybe it would be worth referring to draft-king-irtf-challenges-in-routing: many of the things listed in that draft will be well-known to you, but it will provide a checklist of things to think about (such as implications for scaling in the IGPs and consequences for convergence, etc.) [Andrew AD] have the authors look into Line rate forwarding in today’s routers? [David] I did look at some of the silicons, and it varies a lot. we don’t have actual data set. If anyone has insight, please let us know. [Deren Cisco] depending on the hardware, it is doable. [Acee] take the source for forwarding. It is doable. [Jen] like traditional forwarding, we look at the destination first, looking for more specific destination. then we look at the source. [ZhenBin Li] ACL and Flowspec can satisfy the requirement, not only the source address, it may also include port info. I don’t see the difference. [David] you don’t use ACL or flowspec to build routing table, correct? [Jen] Let’s take it to the list. =============================================== Introduction of a software framework for testing fat-tree routing protocol implementations. Sibyl: a Framework for Evaluating the Implementation of Routing Protocols in Fat-Trees Tommaso Caiazzi (15 mins) https://ieeexplore.ieee.org/document/9789876 [Tony P] I highly recommend people look into their implementation. this lead us to not only prove the correctness of protocol, but also the efficiency. This kind of framework allows us to observe distributed systems in a very formal way, and can help us to validate the protocol extension. [Pascal Thubert] I’m interested to know whether the negative disaggregation really worked? [Tommaso] we didn’t test the negative disaggregation in RIFT. [TianJi Jiang China Mobile] for cloud dc, there are interconnection among multiple sites, and it seems that you didn’t consider like port channels. Have you considered fault tolerance mechanisms that were already put in? [Jeff Tantsura] please continue this work and look at more topologies. [Yingzhen] This is not a draft based presentation. If you have questions, please contact them directly or email the list. =============================================== Presentation 6-11 are new individual drafts, looking for community feedbacks for future work. SRv6 Egress Protection in Multi-home Scenario Wenying Jiang / Weiqiang Cheng (10 mins) https://datatracker.ietf.org/doc/draft-cheng-rtgwg-srv6-multihome-egress-protection/ [Aihua Liu] what is the improvement? did you evaluate with running code? [Weiqiang Cheng - co-author] We have the prototype, the performance of the protection switching is very good. SRv6 Midpoint Protection Zhibo Hu (10 mins) https://datatracker.ietf.org/doc/draft-chen-rtgwg-srv6-midpoint-protection/ [Loa Andersson] Why is this in RTGWG? [Yingzhen] Please take this to the list Usecases of SRv6 Based Computing Interconnection Network Feng Yang (10 mins) https://datatracker.ietf.org/doc/draft-zhang-rtgwg-srv6-computing-connect-usecases/ [Linda Dunbar] Your slides talk about different compute resources requiring different paths. Is the work related to China Mobile’s CAN initiative? [Feng Yang] No, it is not related. [Linda Dunbar] so you’re not considering advertising computing resources? [Feng Yang] no, that’s different. [Tony Li] We already have ubiquitous cloud computing with best effort services. What is the problem you are solving? Why? [Feng Yang] We want to reserve resources for different applications. [Tony Li] What breaks if you use best effort? [Feng Yang] We cannot provide assured service to customers [Tony Li] Sounds like a provisioning problem [Feng Yang] Some of the access network is some form of a tree, so there may be congestion at the tree root. [Tony Li] Sounds like an architectural issue [Feng Yang] Cost of building full mesh is too high [Tony Li] Has been done for many decades [Feng Yang] We want to solve the problem using existing solutions [Lin Han] how do you define the computing resources? name or port? [Feng Yang] it can be central cloud or edge cloud. [Daniel Huang] back to question asked by Linda. Strong recommend to link the work to Compute Aware networking Problems and Requirements of Satellite Constellation for Internet Tianji Jiang (10 mins) https://datatracker.ietf.org/doc/draft-lhan-problems-requirements-satellite-net/ This topic was presented at IETF 112 and 113: Problems and Requirements of Satellite Constellation for Internet Satellite Semantic Addressing for Satellite Constellation https://datatracker.ietf.org/doc/draft-lhan-problems-requirements-satellite-net/ https://datatracker.ietf.org/doc/draft-lhan-satellite-semantic-addressing/ https://datatracker.ietf.org/doc/draft-lhan-satellite-instructive-routing/ [Dino] are all the satellite nodes in one IGP domain? [Tianji] No. it is only for illustration purpose [Lin Han] The traditional technology requires us to configure BGP somewhere to interwork with the Internet, and that’s a problem. [Dino] do you believe GPS based routing will be used for satellite? [Lin Han] there are many options. We are considering them. [Dino] Do you have to interworking with these different solutions with different addressing? [Lin Han] Different satellite networks don’t interoperate with each other. We are only expect one satellite network to interoperate with each other [Dino] so to summarize, each satellite network can just work like a L2 system, then connect to the Internet. [Lin Han] that’s what we expect now. Use Cases for Computing-aware Software-Defined Wide Area Network(SD-WAN) Use Cases of Computing-aware Service Function Chaining (SFC) Shuai Zhang (15 mins) No time for discussion. ## Not enough time to get to this presentation. BGP Blockchain Mike McBride (10 mins) https://datatracker.ietf.org/doc/draft-mcbride-rtgwg-bgp-blockchain/ ===============================================