dinrg at IETF 121 Dublin

• Date: 2024-11-06 13:00 (GMT)
• Chairs: Dirk Kutscher, Lixia Zhang

• Notetaker(s): Dave Plonka, Ryo Yanagida

• 82 participants signed-in to session

• Meeting material

1 DINRG Chairs’ Presentation: Status, Updates

Speakers: Chairs

• Chairs encourage use of DINRG Wiki to share content for group in a
  timely manner: https://wiki.ietf.org/group/dinrg

• Chairs mentioned upcoming DIN workshop at CoNext, 9 Dec 2024:
  https://conferences.sigcomm.org/co-next/2024/#!/din

• see agenda here for the following:
  https://datatracker.ietf.org/meeting/121/materials/agenda-121-dinrg

2 Distributing DDoS Analytics among ASes

• Speaker: Daniel Wagner
• Paper
• Presentation

Discussion

• Yogesh Deshpande: You are multiplying the reflectors, what if the
  reflectors, attacked themselves?

  • Daniel: I am multiplying the reflectors?
  • Yogesh: If you decentralise the reflectors, that brings more
    reflectors closer to the endpoint?
  • Daniel: We do not move the reflectors ourselves. [TODO: Revise
    recording]

• Shane Kerr: It seems counter intuitive that lower bitrate had a
  higher detection rate than higher bitrate.

  • Daniel: Is it this slie?
  • Shane: hmm will come back to it. Next Q. what did you to detect
    falce positives?
  • Daniel: We carried out prominant componence analysis [TODO:
    Revise recording]
  • Dirk: Please take the rest of the questions to the ML

• Lixia Zhang: Given DINRG's focus, Daniel, how does this foster
  collaboration?

  • Daniel: unfortunately, we have nothing operational like this
    (presented) so far.
  • Lixia: How can we grow the distributed the mitigation? What
    roadblocks are challenging this?
  • Daniel: That is more political, than technical, IMO.
  • Lixia: agreed.

3 The Role of DNS names in the Internet Decentralization

• Speaker: Tianyuan Yu
• Presentation

Discussion

• Jens Finkhäuser: Commenting as an individual. DNS is hierarchical,
  any registrar can take the name away from you. i.e. that is
  centralised control at registrar level. You were talking about cloud
  context, again, this is still somewhat centralised. Actually, this
  might be a question. What's the actual difference to what we have
  today?

  • Tianyuan: A subnamespace can be delegated to you. Domain
    namespaces are independent. Personally, I don't think DNS is
    (necessarily) centralized. Users are in control.
  • Jens: In context of cloud service, I could be running a service
    with 1000 users, and I don't see that changing. Now,
    decentralized 1000 users might be better but these identifiers
    are still coupled to a thing that is under someone else's
    control.
  • Lixia: I'll try to answer this too. The decentralisation is
    about TLDs and how many we have. The names do not have to be
    from the service I want to use like github.
  • Jens: That, I agree.

• Niall O'Reilly: What's the compelling case for using the DNS for
  this? We have others (Nextcloud, etc.).

  • Tianyuan: if the 'name' is from the service, you are not in
    control, but your DNS name can be controlled by youself.
    Difference is in control. Not sure if this answers the question.
  • Niall: let's take this to ML

4 Taxonomy of Internet Consolidation & Effects of Internet Consolidation

• Speaker: Marc McFadden
• Presentation

Discussion

• Dirk: on the question of adoption, we have to follow-up on the
  mailing list. Marc, what do you need from the group specifically?

  • Marc: Want to know if participants feel we've made the case re:
    economic consolidation argument?
  • Marc: also are there missing references? Feedback welcome. ...
    are (1) these 4 categories of centralization right? (2) FIXME
    see section 2.1

• Rahul Gupta: We need to consider both software and hardware
  consolidation. Consider "maleable software" (to counter this.)
  Application themselves, like web apps have the power to centralise.
  I think there should be more discussions around that.

  • Mark: Appreciated. In the taxonomy draft, 4th category is indeed
    applications and services. It is a type/point of centralisation.
    None of the drafts here are prescriptive, we did receive
    feedback to clarify more on this.

• Geoff Huston (on zulip chat): I think the case that there is more
  here than economics driving centralisation is not really well made
  (i.e. I'm still unconvinced!)

  • Lixia: I agree with Geoff's comment on chat (zulip). Maybe the
    drive of centralisation isn't always the money.

• Jim Reid: What's the next step? Maybe we need more
  multi-disciplinary approach to this sort of problems. There are
  incentives for security considerations.

  • Mark: I think that is part of the agenda. Maybe IETF can require
    a section considering these aspects. It would be difficult to
    get the consensus but this is probably worth considering.

• Dirk: We re-opend the queue to continue discusion

• Arnaud Taddei: THere are add'l considerations, esp. the Digital
  Opression or resiliency act of the EU. Would it be a good idea to
  study this text (timely, given upcoming event re: this on 26 Jan
  2025). When we say concentration (centralization?) how does it
  overlap with this.

  • Marc: That's a great idea. The intersection betw. consolidation,
    centralization and regulatory controls in place is fuzzy. We
    can't merely regulate ourselves out of this problem. However,
    others believe that is what we should be doing, i.e., protect
    markets. That is a draft for someone else to do (not me.)

• Colin Perkins: As an individual [i.e. not as the IRTF chair].
  Discinction between economic aspects and centralised control etc
  seems fundamental. I wonder how we could answer this. Back in
  sanfrancisco, we had an economist who offered a draft on this.

  • Mark: We should continue our outreach to gather more diverse
    people in to this work.
  • Colin: Agreed

• Jens Finkhäuser: possible answer to Colin, comment is, while I
  haven't read the draft, the shift from economics to power may be
  related or transfer to the work in the Human Rights RG.

  • Marc: if you can read the drafts, send text.
  • Jens: will do.

• Rahul Gupta: We should bring in others (economists, regulators,
  anthropologists) to give us their skills and input on the
  decentralization issue.

  • Marc: There is a lot of academic research, but not specifically
    about the Internet for us to use on this problem.
  • Lixia: agrees, has heard (from academics) that it is great we
    are focusing on Internet decentralization. (It is not in the
    prior literature.)

• Dirk: Please read the draft and think about whether to adopt this

5 DINRG – Next Steps

• Speakers: Chairs & Attendees

• Presentation

• Chairs summarized (in slides) what has been discussed on the mailing
  list.

• Chairs asked: What is the best, realistic role that DINRG can play?

Discussion

• Rahul: Do we have the resources and individuals to take this work
  forward?

  • Dirk: Depends on what people are willing to work on. For the
    academics involved, there is ongoing work to address the
    software dev issues. It will not be the main focus of this
    group; there needs to be activities outside, e.g., research and
    open source. We can't do this all by ourselves.

• Geoff: That summary is one sided. It's the results, not the causes.
  It doesn't provide the understanding of how we got here. This
  situation isn't really unique, we see this in other industries.
  Other industries have dealt with the centralization of economics,
  and dealt with it as an unavoidable consequence, e.g., airline
  industry. We accept this as inevitable in other areas. In the
  Internet, this isn't about resource, but issue around rapid changes.
  Only a small number of players in the industry kept up with the
  rapid chages. This summary only talks about the effects.

  • Lixia: to Geoff, Internet did not start with player(s) having a
    big pile of money. Decentralization needs to take a different
    motivation.

• Juan Caballero a.k.a. Bumblefudge: (an announcement) There will be a
  side meeting, Thursday at 1pm, on the multiformats working group
  proposal.

6 Wrap-up & Buffer

Speakers: Chairs