Skip to main content

Minutes IETF110: spring
minutes-110-spring-00

Meeting Minutes Source Packet Routing in Networking (spring) WG
Title Minutes IETF110: spring
State Active
Other versions plain text
Last updated 2021-04-01

minutes-110-spring-00
SPRING WG - Source Packet Routing in Networking

    Chairs:
        Bruno Decraene <bruno.decraene@orange.com>
        James Guichard <james.n.guichard@futurewei.com>
        Joel Halpern <jmh.direct@joelhalpern.com>

    Secretary:
        Shuping Peng <pengshuping@huawei.com>

===============================================================================
Session I
Monday, 13:00-15:00, March 8, 2021 (UTC+1)

# SPRING Status [ 10 minutes ]
    Chairs

# Compressed SRv6 SID List Analysis [ 15 minutes ]
    draft-srcompdt-spring-compression-analysis-00
    Weiqiang Cheng

# Compressed SRv6 SID List Requirements [ 15 minutes ]
    draft-srcompdt-spring-compression-requirement-04
    Weiqiang Cheng

The two slides are presented one after another.

Keyur Patel: About the analysis draft. On the mailing list, there were some
comparative analysis that are not in the draft. Any reasons? (Chat Darren
Dukes: Keyur, I think you were talking about this earlier?
https://mailarchive.ietf.org/arch/msg/srcomp/ryOuYQbquExcmdi8a9KSNy4FAV4/
      Keyur Patel: @Darren: Yes that is correct)

Weiqiang Cheng: About one month ago, we sent emails to SPRING and 6man WGs for
candidate proposals. But no more feedback. Four proposals are provided by the
WG members. If there are more in the future, we will see how to handle.
Deadline for collecting the proposals for this version was Feb 12th.

Andrew Alston: Simple question. To Chairs and ADs, delays have been caused, but
the developments have been continued and shipping. If we wait forever, it will
be bad for the industry. What is the process? Where to from here?

Joel Halpern: We hope the DT will move. We hear you and do not ignore you.

Xuesong Geng: Why empty tables in the drafts?

Joel Halpern: Have explained.

Weiqiang Cheng: We decided to analyze the four proposals. We will fill in them
soon. We are confident to deliver the full version on the late May.

Boris Khasanov: Different vendors use different chips. How to perform in an
unified way?

Weiqiang Cheng: Basic functionality requirements are done here. No details
about the implementations. Joel Halpern: Need to move on.

# Segment Routing Header encapsulation for Alternate Marking Method     [ 10
minutes ]
    draft-fz-spring-srv6-alt-mark-00
    Giuseppe Fioccola

Joel Halpern: No questions. So we need to move on.

# OAM for Service Programming with Segment Routing      [ 10 minutes ]
    draft-ali-spring-sr-service-programming-oam-03
    Zafar Ali

Joel Halpern: No questions. So we need to move on.

# SRv6 In-situ Active Measurement       [ 10 minutes ]
    draft-song-spring-siam-00
    Haoyu Song

Greg Mirsky: T-flag is at the end of the segment.

Haoyu Song: Not necessary. Can do it in the fast path, but can do it in the
slow path as well.

Greg Mirsky: Need to do it differently in the data plane.

Haoyu Song: We use dedidated probe packets. UDP port is used to indicate the
type of the packets

Greg Mirsky: But you need additional computational overhead.

Haoyu Song: The same as iOAM. Here we can separate the probes from the user
traffic.

Greg Mirsky: Processing in the fast pipeline... Let's take it to the list.

Joel Halpern: No other questions. We move on.

# Performance Measurement Using Simple TWAMP (STAMP) for Segment Routing
Networks       [ 5 minutes ]
    draft-gandhi-spring-stamp-srpm-05
    Rakesh Gandhi

Gyan Mishra: Standards track to Informational, which is good, since it is more
about procedure. Has it been updated in version 5?

Rakesh Gandhi: Yes, we have addressed your comments. Thank you!

Joel Halpern: Please go ahead.

# Enhanced Performance and Liveness Monitoring in Segment Routing Networks     
[ 5 minutes ]
    draft-gandhi-spring-sr-enhanced-plm-04
    Rakesh Gandhi

Joel Halpern: No questions. Discuss on the list.

# Segment Routed Time Sensitive Networking      [ 10 minutes ]
    draft-stein-srtsn-00
    Yaakov (J) Stein

Gyan Mishra: Does this draft update SRv6?

Yaakov Stein: I use segment routing in the title because of the use of stacks
without going into format details. Need to discuss if/how to better integrate
the two.

Xuesong Geng: Similar idea (use of deadlines) was proposed about two years ago.
Interesting idea. Most difficult part is the queue management. This topic
should be discussed more, especially between the protocol and hardware.

Yaakov Stein: This draft is about the use of stacks in the headers, and only
incidentally deals with the forwarder mechanisms. Not sure that theoretical
comparative analysis of scheduling mechanisms is a topic for the IETF. Happy to
discuss. Maybe scheduling strategies can be the subject of another draft.

Joel Halpern: We can move on now. Previous presenters stayed on time so there
is enough time.

# YANG data model for BGP Segment Routing TE Extensions [ 10 minutes ]
    draft-deevi-idr-bgp-srte-yang-01
    Kamran Raza Presented first and then Krishna Deevi

Joel Halpern: No questions. Floor open. Any comments? No.
Thank you!

Speaker Shuffling Time/Buffer:      10 minutes
Total Presentation Time:              120 minutes

===============================================================================
Session II
Thursday, 13:00-15:00, March 11, 2021 (UTC+1)

# SPRING Status [ 5 minutes ]
    Chairs

# SR Replication Segment for Multi-point Service Delivery       [  10 minutes ]
    draft-ietf-spring-sr-replication-segment-04
    Daniel Voyer

No questions.

# SR-TE Path Midpoint Protection        [ 10 minutes ]
    draft-hu-spring-segment-routing-proxy-forwarding-13
    Huaimo Chen

Shraddha Hegde: Are you suggesting that draft should be changed to STD?

Huaimo Chen: No, it is not. Some nodes do not support.

Shraddha Hegde: The solution is to make the nodes to support. There will be a
lot of operations.

Huaimo Chen: Not in that way.

Shraddha Hegde: If B doesnt support, it is the same problem?

Huaimo Chen: We know which nodes are capable. We can send to the nodes that are
capable.

Joel Halpern: We need discussions in the mailing list.

Bruno Decraene: Comment on terminology: your proposal runs during IGP
convergence. It’s more restoration/convergence than “protection”

Huaimo Chen: For the existing drafts, nodes don't know the capabilities of each
other. With the new proposal, we know the node capabilities, then we can choose
the path.

Joel Halpern: Any new improvements you want to include.

Huaimo Chen: Protect the binding segments on a node.

Joel Halpern: Let's move on.

# Segment Routing for Redundancy Protection     [ 10 minutes ]
    draft-geng-spring-sr-redundancy-protection-02
    Fan Yang/Xuesong Geng

Greg Mirsky: In the Detnet WG, there is already a function there.

Fan Yang: We are focusing the replication and elimination functions.

Joel Halpern: Given the overlap the functions, it has to be coordinated.

Greg Mirsky: That is why I asked and made in the mailing list.

Regarding the complexity, 1 plus 1 might be simpler. Here you are doing the
packet by packet, which will bring more complexity.

Fan Yang: They are different mechanisms. More discussions in the mailing list.

Ron Bonica: This proposal gets merging node stateful.

Fan Yang: What kind of state?

Ron Bonica: It sees whether there is a copy of it.

Fan Yang: Sliding window?

Joel Halpern: Goes to the mailing list.

Mach Chen: Controller can be used to configure the merging node.

Fan Yang: This draft only covers the data plane.

Greg Mirsky: Regarding the plan on MPLS, point to Detnet, already provide the
functions. Better to discuss in the Detnet WG.

Xuesong Geng: Already consider the ordering function. For the simple case, the
ordering does not need to be included. 1plus1 protection. This provides
ultra-high protection.

Greg Mirsky: I agree packet duplication minimise the packet loss but with a
cost. Compare the cost and the benefit.

Joel Halpern: We are over time. Please go to the mailing list.

# Seamless SR Problem Statement [ 10 minutes ]
    draft-hegde-spring-mpls-seamless-sr-05
    Shraddha Hegde

Daniel Voyer: There is another draft on SR-MPLS and SRv6 interworking. It is
better to avoid another design team.

Shraddha Hegde: Let's first close the requirement.

Joel Halpern: The two drafts mentioned by Daniel are the following two in the
Agenda.

Cheng Li: Any implementation or deployment?

Shraddha Hegde: Have a solution draft. Have implementation and testing in the
lab.

Jie Dong: Support SRv6?

Sharddha Hegde: Both SRv6 and SR-MPLS.

Jie Dong: Better to change the title of the draft (now it is
draft-hegde-spring-mpls-seamless-sr)

Sharddha Hegde: It covers both already.

# BGP Color-Aware Routing Problem Statement     [ 10 minutes ]
    draft-dskc-bess-bgp-car-problem-statement-01
    Dhananjaya Rao

Tarek Saad: Colar has the same meaning when the packets go across multiple
domain.

Dhananjaya Rao: Key aspect. Clearly define the color domain.
Jim Guichard: Covered in the solution draft.

Linda Dunbar: Difference between your color and the route target?

Dhananjaya Rao: Color is used to express the intent.

Linda Dunbar: Another instructure on how to signal the color.

Dhananjaya Rao: There is the solution draft on how it is used and signaled.

Ran Chen: We have a draft solve the same problem.

Joel Halperner: better to send to the mailing list.

# SRv6 and MPLS interworking    [ 10 minutes ]
    draft-agrawal-spring-srv6-mpls-interworking-05
    Swadesh Agrawal

Dhruv Dhody (PCE co-chair hat on): Talk to the Binding SID authors in PCE WG.
This usecase seems to be allowed, but make it explicit. PCE WG also deals with
Multi-domain case, consider adding that.

Swadesh Agrawal: Will do according to your suggestions.

Daniel Voyer: Here we are also trying to define a solution. Need to collaborate
and cooperate in order to avoid overlapping.

# The SRv6 END.DTM Endpoint Behavior    [ 10 minutes ]
    draft-bonica-spring-srv6-end-dtm-04
    Ron Bonica

Swadesh: End.BM?

Ron Bonica: Very close.

Swadesh: Lookup will give you the MPLS stack.

Ron Bonica: Very close. Maybe lookup is needed.

Swadesh: To me it is End.BM.

Ron Bonica: Very close.

Daniel Voyer: Propose a new end function.

Ron Bonica: Same comment. Can we looking into merge?

Cheng Li: ... G-SRv6 defines the same thing. We can go to the mailing list.

# Building blocks for Slicing in Segment Routing Network        [ 10 minutes ]
    draft-ali-spring-network-slicing-building-blocks-04
    Zafar Ali

Greg Mirsky: Which definition of the network slice is this refering to? Basic
question about the concept of the network slice.

Joel Halpern: The rest of the presentation may clarify.

Jie Dong: First part is a summary of the technologies already defined in this
WG for general purpose. Please refer to the
draft-spring-resource-aware-segemnts and draft-spring-sr-for-enhanced-vpn
drafts. The sharing part will be explained in the presention on
draft-dong-teas-enhanced-vpn-vtn-scalability-02 (happened in TEAS, hopefully
also in SPRING today). Lastly, the slice-ID mechanism is related to IPv6 and
MPLS data plane extensions, which hopefully will be discussed more in 6man and
MPLS WG.

Ali: Agree to work offline.

James Guichard: We need to cut the queue.

Joel Halpern: Quick answers.

Greg Mirsky: Still dont understand which concept of the network slice you are
using.

Cheng Li: A bit is enough for the slice id? Opinion on using the entropy label?

James: Please go to the mailing list.

Tarek Saad: Slice is not independant of the topology.

Joel Halpern: Please go to the mailing list.

Daniiel Voyer: I agree with Robin (Huawei)'s email to the list.

Joel Halpern: Be clear about the definition. Let's move on.

# Scalable Network Slicing over SR Networks     [ 10 minutes ]
    draft-bestbar-spring-scalable-ns-01
    Tarek Saad

Zihbo Hu: The solution of multiple slices sharing the same topology can use
Per-slice SID or the slice ID extension in data plane. This solution has been
proposed in 2019 in draft-dong-teas-enhanced-vpn-vtn-scalability and the
relevant protocol extension drafts, Slice aggregate is the same as VTN, no need
to repeat this work.

Tarek Saad: We are happy to do the alignment if needed.

Jie Dong: As discussed in TEAS, terminology consistency is important. Secondly,
is the SA SID a resource-aware SID?

Joel Halpern: Similarity with the resouce SID. We need to sort it out.

Tarek Saad. We will follow up. No protocol extensions.

Zafar Ali: Backward compatibility. All nodes?

Tarek Saad: Backward compatible. There are multiple ways to carry it. I like a
pointer.

Zafar Ali: Good to see that in the draft.

Joel Halpern: Please go to the mailing list.

If time allows:
(If time does not allow, those two presentations can be attended respectively
in the TEAS and RTGWG WGs)

Due to lack of time, the following two presentations could not be presented.

# Scalability Considerations for Enhanced VPN (VPN+)    [ 10 minutes ]
    draft-dong-teas-enhanced-vpn-vtn-scalability-02
    Jie Dong

# Associated Channel over IPv6  [ 10 minutes ]
    draft-yang-rtgwg-ipv6-associated-channel-00
    Fan Yang

Speaker Shuffling Time/Buffer:      5 minutes
Total Presentation Time:              120 minutes

===============================================================================
Joint Session with MPLS, DETNET, and SPRING
Friday, 13:00-15:00, March 12, 2021 (UTC+1)
Pseudowire And LDP-enabled Services