87th IETF, Berlin, Germany July 31st, 2013 Information-Centric Networking Research Group Session Chairs: Dirk Kutscher, Börje Ohlman Presenters: Kostas Pentikousis, Jan Seedorf, Stefan Lederer, Mark Stapp, Lixia Zhang, Damien Sauze, Elwyn Davies Questions from: Georgious Karagiannis, Univ Twente Ken Chunsuan Xong, Huawei Thomas Schmidt, Hamburg University Yunfei Zhang, Coolpad Jan Seedorf, NEC Kostas Pentikousis, Huawei Technologies Scott Brim, Internet2 Börje: IRTF IPR Policy presented. ICNRG introduced. Dirk: Volunteers for Jabber monitoring? (Elwyn Davies volunteers) Börje: The slides for today are in the Wiki page ( LINK: http://www.ietf.org/proceedings/87/icnrg.html ) . Upcoming meetings presented. Agenda bashing. (1. Update from Sunday) Börje: Pointed out main events from Sunday. Dirk: Remarked on how useful full-day meetings are for discussing and progressing the work. Provided a quick update on the Research Challenges ICN draft ( LINK: http://www.ietf.org/proceedings/87/slides/slides-87-icnrg-7.pdf ). Question (Georgious Karagiannis, Univ Twente): Association between the routing forwarding and caching placement. The draft discusses 2 options: off and on path. Is there any possibility of an hybrid solution where you place the cache only on some routers? Is this an option to be studied? Dirk: We should not need caches in every element of the network. Börje: You have to change state for every packet, but don't need to cache it always. Dirk: Continued with the update on the changes to the draft. Question (Georgious): Are dynamic objects associated only to data objects, or with functions as well? Börje: Also functions. Georgious Karagiannis: I propose another domain of study, cloud computing, associated to Network Virtualized Functions. You could use ICN to copy certain virtualized functions and distribute it for different locations. Dirk: Would be interesting. But isn't it more an application of using ICN? Börje: This could be discussed on the mailing list. Kostas will now discuss the ICN Baseline Scenarios draft. Kostas: Provided an update on the Baseline Scenarios draft. (around 10 people acknowledged reading the version of the draft in Orlando, and a different set of 10 people acknowledged having read the latest version of the draft). The goals of the draft were described. Georgious Karagiannis: We can provide information on the NFV scenarios we mentioned. (2. Using ICN in disaster scenarios, Jan Seedorf) Jan: Slides presented ( LINK: http://www.ietf.org/proceedings/87/slides/slides-87-icnrg-2.pdf ). Question (Ken): Have you thought about a segmentation between the first responders and the general public? Jan: We have been talking about prioritization. Every router can know if the content is important or not. We haven't talked about it yet. Ken: Have you thought going into the Energy Efficiency Management WG? Jan: We started to look in that. Question (Chunsuan Xong, Huawei): Have you considered protection mechanisms about fake information? Jan: Its a key point. Its looked on in the draft. Question (Thomas Schmidt, Hamburg University): Bursty traffic is primarily conversational. The strength of CCN is more content oriented. Can you comment? Jan: A lot of Japanese partners gathered a lot of data from the earthquake that has happened. For that you can imagine that you publish a message for your family that you are alive, so it is more Publish/Subscribe mechanisms and that's where we're looking at. Question (Yunfei Zhang, Coolpad): Regarding authentication. Do you consider authentication of the data or the users? Jan: From both. ICN has solutions already for authentication of the data. (3. Adaptive Streaming over ICN, Stefan Lederer) Stefan: Slides presented. ( LINK: http://www.ietf.org/proceedings/87/slides/slides-87-icnrg-5.pptx) Question (Chunsuan Xong): Caching may cause some oscillation. You can use the position of the segment to ask in the cache. I don't know why you have oscilationd. Stefan: It measures also the download speed of the segment, and measures the amount of bandwidth available to me. If the bandwidth is much lower, it starts to oscillate. Chunsuan Xong: It is not caused by the cache. Stefan: We need adaption logics to handle this, we check the buffer in the client not the bandwidth measurement. Question (Jan Seedorf, NEC): The CDNi published a document on adaptive streaming and it explains why it is not a problem. Stefan: I have to disagree, because in CDN the environment is managed. In ICN you have no change to determine. (4. Fast Forwarding for NDN, Mark Stapp) Mark: Slides presented. ( LINK: http://www.ietf.org/proceedings/87/slides/slides-87-icnrg-4.pptx ) Question (Kostas Pentikousis, Huawei) Why do you consider so small packets? Mark: Average. Interest packets only contain name. Kostas: In TCP you see a lot of big packets. Like 60% more and the rest is acknowledgements. Mark: We're trying for a conservative number. (5. NDN Simulator, Lixia Zhang) Lixia: Slides Presented. ( LINK: http://www.ietf.org/proceedings/87/slides/slides-87-icnrg-10.pdf ) (6. Network Experimental Prog. Interface (NEPI), Damien Sauze) Damien: Slides presented. ( LINK: http://www.ietf.org/proceedings/87/slides/slides-87-icnrg-1.pdf ) Börje: Due to time we need to finish presentations here. (7. Turning Challenges and Scenarios drafts into RG drafts) Comment from Lixia: There is a different between writing and experimentation, that is missing in the drafts. Dirk: Some challenges have a better understanding than others. Kostas: This refers to the Challenges of Scenarios? Lixia: The Challenges. You imagined what were the challenges, which is different from experimental work. Comment (Scott Brim, Internet2): Even if they aren't right now, after they become RG drafts, they will be changed and a lot will be added. Lixia: That is a decision by the group. Börje: The issue is if we move them into WG drafts. Kostas: People are coming from different projects. Nothing prevents having a parallel NDN Challenges draft based on experimental. That does not impede ICN challenges to evolve. Dirk: In general different teams are working in experimentation and we need to encourage them to present their work here. Börje: To me it makes a lot of sense to split the scenario draft into two drafts, coming up with a name with what's now in section 3, which is about evaluation methodology and metrics. Anyone has a problem with that? (silent room) Börje: So, no comments. Can we humm for doing that split? (no strong opinions, but there was significant hum for doing the split) Dirk: We have people working on a survey draft. is there anyone here who can speak on behalf of that? Börje: We basically have 3 drafts to become RG. I would propose to do all three of them RG, or just wait not to name one. You have a view Dirk? Dirk: Let's have the authors decide, not discuss it here. Börje: I think it's ok to name this ICN RG drafts. Because we don't have alternative ones and, of course, we'll continue this. So the proposal is to turn all these threes drafts into RG. Any comments? Can we humm? Who supports doing that? (silent) Börje: Who is against it? (silent) Börje: Then maybe we can come back on this. Dirk: We give the chance for people to object on the ML to this. Börje: If somebody would like to ask people for any ideas for new directions that anyone would like to propose? Any new drafts we would like to start? Börje: One proposal is this Video Distribution is turning into quite an interesting topic, it could be a potential new topic. Can we have a hands check for interest in this draft? (some hands raised) (8. Honk Kong agenda) Börje: We will have a full meeting day there.