ICNRG
IETF104
Friday March 29 2019 - 10:50 to 12:50 - Room Karlin I/II aka the room with the monolith
Chairs Intro: Agenda Bashing, Minutes taker, Blue sheets, Status (5 min)
https://datatracker.ietf.org/doc/slides-104-icnrg-chairs-slides/00/
there has not been a lot of contributions
The minutes are here:
https://conferences.sigcomm.org/acm-icn/2019/
ICN conference to be held in Hong Kong 24-26 September with ICNRG interim a day before so plan to arrive early.
Broader CFP.with added ancillary topics e.g. distributed network, metrics, use cases, applications, deployment etc.
Should strengthen the program.
May 3 paper registration, May 10 paper submission (hard dates no extensions).
HopAuth (Hop-by-Hop Authentication with Suspension Chain Model) - Ruidong Li (15 mins)
https://datatracker.ietf.org/doc/slides-104-icnrg-hopauth-ruidong-li/00/
Good description of the security mechanisms in the internet in the introduction then description on how it can be done in the ICN world.
Draft needed?
Question (MJM): What is the link between this approach and distributed authentication based on the blockchains? Are both needed?
Answer: they are different and use different mechanisms and have different targets - we target the routers blockchains are more at the application layer - both could be used.
Question: (DaveO) what do you lose if all hops do not implement the authentication?
Answer: we have implemented a hashing mechanism to address this
Comment: (DaveO): slide 10 on forwarding - CCN not evict authenticated content to replace it by non-auth - not cache poisoning but can permit cache pollution
ICNLowPAN Update - Cenk Gündoğan (15 mins)
https://datatracker.ietf.org/meeting/104/materials/slides-104-icnrg-icnlopan-update-cenk-gundogan-00
https://tools.ietf.org/id/draft-irtf-icnrg-icnlowpan-00.html
Question: (DaveO) re: Time TLV slide. Is there a reason we should not deprecate the previous TV for Interest Lifetime?; no downside to use this as the way to manage TLV vs. other proposed approaches
Answer: Only issue is that here you have to do computations not just use bits directly.
Comment: (DaveO) most cycles go into data structure for expirations (e.g. clock queue) - parsing isn’t the issue
LowPan reduces the message size and the time in flight pf the packet; reduces the energy consumption. All outweigh the compression complexity.
Question: (Dirk) on requests from the author. Is there more work needed?
Answer: Pretty mature work.
Comment: (Dirk) NDN/CCN expert need to review the work; volunteers wanted
Comment: (Thomas) Discussion needed/continuing on the list on time development TLV (support of DaveO’s comment)
Comment: (DaveO) We need the reissue the CCNx messaging with the new TVL. We can deprecate the old TVL - for CCNx not NDN yet. But will take the action to talk to the NDN people.
Enabling ICN in 3GPP’s 5G NextGen Core Architecture Update - Debashish Purkayastha (15 Mins)
https://datatracker.ietf.org/doc/draft-ravi-icnrg-5gc-icn/
Comment: (DaveO) Please read the draft and comment. We need to decide if we want to accept it as a RG draft.
Finalizing the Terminology Draft - Bastiaan Wissingh (10 mins) presented by Dave Oran.
draft-irtf-icnrg-terminology-03
(short presentation)
DaveO: Will poll the list through an RG last call and send to IIRSG Review if no problems
QoS Treatments - Anil Jangam [remote] (20 mins)
https://datatracker.ietf.org/doc/slides-104-icnrg-qos-treatments-for-icn/00/
https://tools.ietf.org/html/draft-anilj-icnrg-dnc-qos-icn-00 (is there an update?)
Intro by DaveO:
Little in depth work done on QoS differentiation of ICN traffic. DaveO and others presented in misc. interims. We now have 2 drafts.
Goals for today:
Take this as final kick-off of the QoS work.
Question: (Cenk Gündoğan) re: slide on PIT enhancements: Table of QoS markers or any markers work; could you have malicious use of QoS markers.
Answer: This table is a placeholder - markers mean nothing at this point.
Comment: (DaveO) Communicated to the authors that is an unreasonable Pending Interest Table (PIT) design. But there could be a valid scalable design and the authors are aware of this.
QoS for Constrained Environments - Cenk Gündoğan (15 mins)
https://datatracker.ietf.org/meeting/104/materials/slides-104-icnrg-icn-qos-for-iot-cenk-gundogan-00
https://tools.ietf.org/html/draft-gundogan-icnrg-iotqos-00
RIOT+CCN-lite and testbed coming.
Comment (Thomas): Late draft is because of QOS in ICN more interesting than IP advantage is correlating these concepts
Comment: (DaveO) tradeoff; need to infer classes in the forwarders a priori you need to configure it; EC3 (EC3=equivalence class component count) scheme could also be a solution if you do not have to have QoS on the 1st packets. You may have the best of both world here.
Refresher on Flow Classification [tentative] - Dave Oran (15 mins)
? https://tools.ietf.org/id/draft-moiseenko-icnrg-flowclass-02.html
Review of the flow classification draft.
EC3=equivalence class component count
ECNCT= equivalence class name component type
Should we adopt the draft? DO both?
Comment: (Börje) We do QoS to reduce the traffic; putting markers in the name is risky and could increase the traffic. Avoid sending traffic in many versions and we should have a discussion on this at one
Answer: QoS treatment or EC? Markers w/o limitation is how you use then can have a deleterious effect.
Question: (Cenk Gündoğan) Re: ECNCT (already discussed on the list) new name component type on CCNx there is 2 bytes of space for name component would the app be about to use other type
Answer: this is not what is proposed; is does not prevent the type (long explanation/implementation follows)
Question: (Cenk Gündoğan) In NDN you can request data with a prefix.
Answer: prefix mac on data is bad idea - but good point: how ECNCT interacts with prefix matching needs attention to support NDN
Comment (Dirk) The is a way to do QoS much better; any objection to adoption? Will ask on the mailing list if we adopt this draft.
EU Open Call for Next Gen Internet - Brook Schofield (sp?) (5 mins)
Chairs Wrapup: (1 min)