IABOpen, IETF 110
Thursday, March 11, 2021, 15:30-16:30 UTC+1
Chairs: Mirja Kühlewind, Zhenbin Li
Welcome and Status Update
What is IABopen?
IAB organized session focusing on technical and architectural topics to:
architecture-discuss@iab.org for architectural discussions and IAB documents
iab@iab.org for comments and concerns directly to the IAB
Program mailing lists for direct comments on work of a specific program
Document Updates
RFC 8980: Report from the IAB Workshop on Design Expectations vs.
Deployment Reality in Protocol Development
draft-iab-covid19-workshop-01: Report from the IAB COVID-19 Network Impacts Workshop 2020
After the restructuring of IAB Programs into Technical Programs and Admin Support Groups, the IAB is now reviewing all programs.
IAB IETF-IANA Group
Liaison Oversight Program closed
The IAB is in the process of reviewing its liaison management function.
The IAB has established a new Liaison Coordinator position.
The IAB is working actively towards further improvements.
Éric Vyncke: This is a nice change, is it public somewhere?
Mirja Kühlewind: It’s in the IAB report about the liaison coordinator position, but it’s all fresh news and the website will be updated soon.
John Klensin: I want to say that I think this is a step in the positive direction, and I really appreciate that the IAB is taking this seriously again and moving in a direction to keep everyone more informed and coordinates.
Mirja Kühlewind: Definitely, that is the goal here.
Charles Eckel: One of the things I have been spending a lot of time on is open source, and there is a nice ongoing set of talks that Open Forum Europe is putting on. I also work closely with the Linux Foundation. I don’t know if a liaison is the right thing, but something like that for open source, as the worlds of open source and standards continue to affect each other.
Mirja Kühlewind: In the cases where we had to establish a liaison relationship, it is often because the other organization requires a formal relationship to see their docs. The other point is to figure out where we need to go and who we need to talk to, and I see that as independent of liaison management itself, it’s more about outreach. It’s definitely something I want to drive forward, but it’s less straightforward. More discussion is needed.
Charles Eckel: There is an optic component, because it’s not obvious that there are discussions happening when there’s not a formal liaison.
Wes Hardaker: All technical discussions, there is nothing stopping IETF participants from engaging in other communities. The IAB would like to hear the places where that’s not working and where we need more formal relationships. But don’t let any of this program stop people from working in multiple locations and bringing information back and forth.
Mirja Kühlewind: If you have information you think is important, please send to the IAB.
Robert Moscowitz: Last year I got involved with ANSI on aircraft systems and had to represent the IETF and what we are doing in our standards. I did reach out but only got minimal support. ANSI then spun something up on satellite communications, but I didn’t have the bandwidth to follow. My experience has been a struggle as I deal more with close SDOs that have very formal rules on liaisons. Having more IETF backing may be of real value. There are other areas where we could do better, and I don’t know how to proceed on this.
Mirja Kühlewind: I understand when you say need more backing form the IETF, but we still would need someone like you who has the bandwidth to go there and participate. It’s a much stronger commitment.
Bron Gondwana: I’ve ended up as a bit of an unofficial liaison in a couple of places. Does this need to be officially tracked?
Mirja Kühlewind: There is no requirement to track that. It’s good if people keep an eye out and promote IETF work in other places. You’re not officially speaking as the IETF, but if they need someone who can, please reach out.
Simon Hicks: I am vice chair of the ETSI board, but I am not speaking for them. It works awkwardly in the reverse direction as well, it’s awkward for ETSI to figure out to talk to unless we have individuals participate. I think IETF would benefit in the overall SDO ecosystem if there was a central contact for this, perhaps in the Secretariat, although I realize that is not how the IETF Secretariat work. But we need to have ways to talk to each other or it will get more awjwards and fragment the standards ecosystem.
Mirja Kühlewind: The official contact point is the IAB, and we hope having the liaison coordination contact will help with this. If you use any of the other email addresses on the IETF website, they go to the Secretariat and they know how to reach the IAB.
Welcome to David Schinazi as co-lead!
Status Update
Protocol Extensions
We’re still learning
Get involved!
Eliot Lear: One quick comment, I think there is difference between protocols and formats. When you look at each, with a protocol you can negotiate and fail fast. With an object format, you have less opportunity to fail, so you have to look at it differently, like how to handle backward compatibility.
Slides: Architectural Concern in Focus: Centralization (Personal Viewpoint and Pointers, Jari Arkko)
Mirja Kühlewind: Deployment is driven by the benefits a solution provides, price, ease of deployment, and these are things we should keep in mind when we design our protocols.
Eliot Lear: This is good work and I think it’s something that has been on people’s minds for a while, and I want to acknowledge early comments Mark Nottingham made on this in an IAB plenary way back in 2014 or so. I do think there are times when we should be thinking about these aspects architecturally. One of the most egregious aspects of private networking is that it forces concentration. This is a problem that can recur. Martin has written on the risk of intermediaries, which comes into this discussion. And I want to agree about how complex a problem this is. Mail services are considerably more reliable than there has ever been because there is a concentration of services and expertise, but when one of these fails then it’s a problem. One thing we should be thinking about is, can standards play a role?
Pete Resnick: Purely questions. One thing is, where can folks get more involved in the discussion and participate? A lot of people care about his and would like to give input. And, does the IAB have any good levers to pull to deal with this?
Jari Arkko: I think the only lever we have is education. As to where to discuss this, yesterday someone suggested a workshop, but we’ve had workshops on this. Definitely architecture-discuss is a place.
Pete Resnick: I think something more long-lived than a workshop would be good.
Andrew Campling: Thank you for covering this. You touched on that there can be some benefits to users on centralization, but I would argue that is likely to be untrue over the long-term because of privacy etc. I do think there needs to be an ability to review and challenge new protocols before they are allowed as standards to see if they make the problem worse, and maybe intervene to ameliorate them. Otherwise this is a disaster for the end user.
Dominique Lazanski: I support Pete’s idea about doing something that is IAB-sponsored, maybe using a workshop as a jumping off point, but having and ongoing place for it as well.
David Schinazi: I agree that this is a complex topic that is important, and I agree that too much centralization is bad, but I am confused about what the IETF and IAB can do here. We are not the protocol police, we are not legislators. If we write an RFC saying a centralization is bad, that doesn’t move the needle. I think it is out of scope. I think what we can do is make sure our protocols are open and extensible.
Jari Arkko: I think your point is valid, but it’s more pointed advice that here are techniques that can help.
Tommy Pauly: I want to clarify that I agree with what David said, but we should be focusing on promoting protocols that allow distributed and non-centralized ways of doing things.
Martin Duke: I share the concerns of a lot of people have about this. I think the work of creating interoperable standards is inherently decentralizing. In my non-expert opinion, centralization is a legislative and legal problem. We do ADD and things like that, but I have trouble seeing the levers the IETF has to move the ball forward
Eric Rescorla: David and Martin said a number of things I would say. Need to distinguished between things that are inherently centralized and the things that are not. The extent to which you might ask what the IETF can do to make it less centralized, what are the technical functions that can de-incentivize this? Those are the anchors or levers that we have to pull. RFCs or blog posts won’t have much of an impact.
[Session ran long and was already over time before getting to this agenda item.]
Mirja Kühlewind: Feel free to send email to iab@iab.org if you want to talk to us, and I think the consolidation discussion can continue on architecture-discuss@ietf.org.