IETF 120 SRv6OPS WG Rough Minutes

Monday, 22 July 2024
15:30 - 17:00 (Vancouver, Regency A/B)
22:30 - 00:00 (UTC)

• Chairs: Daniel Voyer, Dhruv Dhody, Weiqiang Cheng
• Note Takers: Thomas Graf, Boris Kh, (updated by Dhruv Dhody)
• Slides
• Video
• Chat-log
• Meetecho Recording

Introduction

1.1. Administrivia, Agenda Bashing (Chairs, 5 min) [5/90]

1.2. Charter & Milestones Recap (Chairs, 5 min) [10/90]

1.3. A word from the Responsible AD (Jim Guichard, 5 min) [15/90]

• Jim Guichard: Working with SPRING chairs on a recharter

Operator Presentation

2.1 Verizon: SRv6 in Verizon (Luay Jalil, 15 min) [30/90]

• Mike McBride: In brownfield environment what is best guidance for
  transitioning from MPLS to SRv6?
• Nick Buraglio: Could you describe how you manage risk and perform
  perimeter filtering for SRv6
• Luay Jalil: Will discuss offline and on the mailing list
• Ketan Talaulikar: To mike, there is a draft in SPRING
  undergoing adoption for interworking between SR-MPLS and SRv6
• Tony Li: Just want to point out that there is no reason why it can't
  be run ship in the night with both running in parallel
• Luay Jalil: We have deployed in parallel and encountered that there
  are many issues (like cost) with such an approach.
• Ketan Talaulikar (chat): Ongoing work and adoption call in SPRING:
  https://mailarchive.ietf.org/arch/msg/spring/qVc41lFqOknxAluPGYLjNWbk560/

2.2 China Unicom: SRv6 in the Cloud (Xinxin Yi, 15 min) [45/90]

• Eduard V (chat): Actually, E2E services (from Edge to VM) is the
  primary advantage of SRv6. SR-MPLS could not give E2E.
• Dhruv Dhody (chat): @Xinxin Yi - lets start a thread in the mailing
  list to talk about your deployment! I wanted to undestand if you are
  using NSHoSRv6 or integrated NSH as per RFC 9491
• Xinxin Yi (chat): Of course it's good. NSH has some other issues,
  such as complicated configuration，and SRv6 more convenient.

2.3 Rakuten: SRv6 deployment case study (Akash Agrawal, 10 min)
[55/90]

• Daniel Voyer: (Regarding the graph with 93% reduction) From which
  network design are you coming from that now you have 93% reduction
  of incidents is impressive.
• Akash Agrawal: This ratio comes from comparison wih pure IPv6
  network, since we do not have MPLS.
• A clarification was made that the reduction in convergence time was
  for the transport IP domain.
• Tom Hill: Which feature of SRv6 helped in reduction of network
  incident.
• Akash: In the hop by hop routing architecture, routing consistency
  is needed on every node because routing decision is taking hop by
  hop. The tunneling with underlay and overlay helped.
• Weiqiang Cheng: How many compressed SID in your longest
• Akash: We are using Flex-Algo with one SID.
• Dhruv: Please and come and participate in WG and document the gaps,
  challenges etc
• Akash: One issue we faces was for path MTU discovery
• Alvaro Retana (Chat): FYI about PMTU. The spring WG just finished
  the adoption of
  https://datatracker.ietf.org/doc/draft-peng-spring-pmtu-sr-policy/

Initial Draft Discussion

3.1 SRv6 Deployment and Operation Problem Summary (Yisong Liu, 10 min)
[65/90]
draft-liu-srv6ops-problem-summary-03

• Dhruv: Please limit author list to 5. For the telemetry part think
  about those metrics that might be missing and lets provide inputs.

3.2 IPv6 Address Assignment for SRv6 (Yongqing Zhu, 5 min) [70/90]
draft-liu-srv6ops-sid-address-assignment-00

• Dhruv (from the chat): why sharing the same node SID for P2P and
  P2MP service was an issue in your deployment?
• Dan (from the chat): what's the use case behind that ?

3.3 Best Practices for Protection of SRv6 Networks (Changwang Lin, 10
min) [80/90]
draft-liu-srv6ops-sr-protection-01

• Dhruv: The document list all the various techniques and jumps to
  recommendations but does not give a rationale. Please work on it.

3.4 Alternate Marking deployment status (Tianran Zhou, 10 min) [90/90]

draft-zhou-srv6ops-am-deployment-status-00

• Not presented becasue of audio issue! Will be discussed in Dublin.

Last updated by Dhruv Dhody at 23 July 2:40 PM PT