ICNRG Interim Meeting Notes
November 4 2018 - Bangkok, Thailand
Chairs Intro: Agenda Bashing, Minutes taker, Bluesheets, Status
Minutes taker: Marie-Jose Montpetit
Main Meeting on Thursday
Interim is to work on other topics or give details on existing topics
Material available on the meeting materials site
Report from September ICNRG Interim meeting in Cambridge, MA, USA
(See agenda in previous meetings” on the wiki)
Scuttlebutt as an ICN system; ICN should get more acquainted with the work
Martin Benoit on ICN video over the ENCQOR 5G testbed with blockchains/ledgers for the rights protection
Naming resolution services (later today also) need to cast a wider net coming out of the NSF dressing global issues under Mobility First vs. CCN model. Global name resolution is a research topic. GNS may be more included in ICNRG. Presentation at the interim was an interesting approach. (look at the presentation)
Low-Pan update - to also be presented Thursday
NDIST presentation on trace route
NDN IOT: extensions running on RIOT
QOS discussion (also today)
No questions
Updates to 4G/LTE the draft based on feedback and additional work - Prakash Suthar
https://datatracker.ietf.org/doc/draft-irtf-icnrg-icn-lte-4g/
Comment from the presenter: thanks to the chairs for an outstanding job!
There is also a draft on 5G with Ravi Ravindran that will be presented on Thursday.
Simulation via NS3 LTE and EPC models
Questions: last slide is not related to the 4G and LTE and is more general.
Close the work on the 4G/LTE. Adopted 2 meetings ago and has had 2 iterations. Not controversial.
This is work that needs to be done so that ICN research can work.
Not research but there is no IETF on ICN.
Gray area between informational vs. experimental: bring to the mailing list. Chairs action item.
ICN QoS
Chairs felt that this in important enough to request research on the topic.
Follow up” “What is this all about” presented in September.
"What to do and not to do" - Dave Oran
(repeat of the talk from the September meeting)
Comment (Borje O.): in 3GPP not IP QOS is not a non-resolved issue
Comment (Thomas): difference between IP vs. ICN is the concept of “receiver” and “sender” is different
Comment: Borje: Producers can delay sending a response to an interest when there is congestion;
Answer to the comment (Dave O.): maybe the interest message should have been filtered
(discussion between Dave O. and Thomas)
Comment (Dave O.): ICN as an opportunity to make stride in the QoS world as IP QoS is not making progress anymore
QoS in ICN using Disaggregated Name Component Approach - Prakash Suthar
Not a draft yet - ideas for the future.
Questions/comments:
(Borje): Objects with different QoS would have different names
(Dave O.): You will need different copies because the encrypted versions will be different
(Dave O.) need a tutorial on HTTP: part parameter
(Prakash): will be part of the draft
Discussion on the content name and routing
(Thomas): how do you aggregate?
(Dave O.) Keep resources on the return link upstream 2nd is the resources of the producer itself.
Need to consider the producer resources (missing from Dave O.’s presentation)
Implementation issues to be addressed later
Discussion on interests and differentiating duplicated interests as retransmissions
(Dave O.): Not more flexible but more practical
(Dirk K.): Keep the marker outside the security envelope ( reason not to be part of the content)
(Dave O.): Mutation of the QoS markers is an advantage.
Multiple drafts possible and should be submitted.
(Dave O.): not use one thing for QoS marking and flow classification
(Borje): Agree and should not be in the name - The authors were asked to explain in the mailing list their motivation for putting the QoS marker in the object name.
ICN and Name resolution
Presentation of newly RG adopted NRS drafts
2 RG drafts + 1 individual draft`
Comment/question (Dave O.): There are 2 dimensions : whether the name resolution is before before
requests are injected or after
(Dave O.): you should Contact Arun D. from U. Mass as the implemented GNS for Mobility First.
(Borje): this is generic not NND
(Dave O.) then this can integrate the GNS.
(Dirk): Typo: 2018 not 2019 for the release date
(Dave O.): Use an open source KV store to get away from the license issues.
(Prakash): We need NRS to deploy ICN. I will work with you.
RICE next steps discussion
Named function networking in ICN - assumed both static and dynamic data .
RICE addresses remote invocation - a hard problem.
Paper in ACM ICN (best paper award)
Comments/questions: (Thomas): how does the “call again”
(Dirk): Could be automated.
(Thomas): The is a distributed system of processes.
(Dave O.): If latency not a issue you can give on overestimate.
(Dave O.) 2 aspects: temporal and sharing of caching
(Dave ).): 90% requires protocol changes - need feedback from the community
(Dave O.) Reflexive forwarding could be used for other things
Compute-First Networking and heads-up for COIN
Please come to the COIN meeting on Friday 10:00-12:00.
Chairs Wrap-up
Please come to the regular ICNRG meeting on Friday 9:00-11:00.