Minutes of the IP over Digital Video Broadcast WG (IPDVB) =========================================================== TUESDAY, 13:00-15:00 November 8, 2005 Chair: <gorry@erg.abdn.ac.uk> WG Note-taker: Martin Stiemerling <stiemerling@netlab.nec.de> 1. Agenda --------- (WG Chair, Gorry Fairhurst) This was the 5th meeting of the IPDVB WG at the IETF. The meeting was chaired by Gorry Fairhurst. The proposed agenda was accepted. An additional presentation by Axel Jahn on IP over DVBS.2 had been received, which the WG Chair would present. 2. Document Status ------------------ (WG Chair, Gorry Fairhurst) Active WG Drafts: draft-ietf-ipdvb-ule-06.txt (Proposed Standard) RFC Editor queue in EDIT IANA. draft-ietf-ipdvb-arch-04.txt (Informational) RFC Editor's queue in Auth48 call. draft-ietf-ipdvb-ar-01.txt Active WG draft. Protocol entries for ULE has been created in registries at: ISO/SMPTE ATSC IANA Next-Header registry These registry assignments are now permanent. Other related documents Individual Drafts for consideration by WG: draft-stiemerling-ipdvb-config-01.txt draft-cruickshank-ipdvb-sec-req-00.txt draft-cruickshank-ipdvb-sec-00.txt draft-cantillo-ipdvb-s2encaps-01.txt There had been a re-charter of the WG Charter milestones, after the last IETF meeting. A new item added on ULE security requirements. The new milestones are progressing well and all completed on time. 3. Presentation on IP Address Resolution ---------------------------------------- (draft-ietf-ipdvb-ar-01.txt) Marie-Jose Montpetit Marie-Jose presented an overview of the document structure and noted that this must work in all MPEG-2 networks: address resolution should be technology agnostic. The I-D was to: Define terminology, lead to a common understanding, review implementation scenarios. Address resolution should also be considered above IP and the goal is how to make the implementation more IP friendly. The I-D is applicable to whatever MPEG-2 transmission is used (satellite, cable, handheld, ...) and presents methods to ensure AR is network-aware / technology agnostic (mobility of solutions) and speaks about integration into multiple signaling paradigms: DVB-SI and IMS/SIP; This can resolve other parameters: encapsulation method, MTU, policy/priority/QoS, security, packing threshold. The current revision -01 is now in good shape. Updates include: Added description of use of SI. Various things based-on discussion on the list. Fixes to document format. She presented the intended work for I-D rev. -02: Investigate integration into other new trends: SIPPING, ipconfig, autoconf; Closer integration into other standards bodies; Inputs are requested from the WG on mobility, DVB-H, IMS; Also: cable/broadcast strategies usage, IP streaming over MPEG-2. Gorry Fairhurst: There are a lot of things that could be added. How long until the I-D is ready for a WGLC? Marie-Jose Montpetit: This is some kind of shopping list, we propose to cycle it one or two more times, and get comments from the list. The I-D should be ready by early next year. Gorry Fairhurst: The milestone is March 2007. Marie-Jose Montpetit: Seems OK. Gorry Fairhurst: I encourages people to read next draft, and offer contributions where able. 4. Presentation on IP Address Configuration ------------------------------------------- (draft-stiemerling-ipdvb-config-02) Martin Stiemerling There has been progress, the I-D has been reorganised, and is now ready for comments from the WG. Open issues include the configuration parameters, sets of configuration items, and how to reconfigure? The I-D still needs to differentiate between the configuration roles of the MPEG-2/DVB operator and IP operator. There is no current conclusion on whether new protocols are needed? Marie-Jose Montpetit: The I-D speaks about MPEG-2/DVB and IP operators. What about content provider configuration? Martin Stiemerling: That's a good question. It should enter the picture somewhere. Marie-Jose Montpetit: All these people have configuration. Martin Stiemerling: If we talk about IP-level questions, then I am not sure they are involved, if this is at the service-level, then there is configuration. Gorry Fairhurst: Does anyone have NDP experience in very flat networks - i.e. scalability to large numbers of systems? Marie-Jose Montpetit: The Secure Neighbor Discovery WG (send) has work on scaling, captured in an RFC. This is important to the AR I-D. Martin Stiemerling: I will look into this point. Gorry Fairhurst: Is the I-D heading towards an Informational RFC. Should the outcome be part of another I-D? Martin Stiemerling: There can be recommendations on DHCP, NDP, arp, etc. Gorry Fairhurst: This seems to fall within the Chartered item on Address Resolution. We may finally need to consider whether this I-D should be combined with the AR draft. For now, I am very happy to see the I-D proceed, and would like to see contributions. 5. Presentation Security Requirements ------------------------------------- (draft-cruickshank-ipdvb-sec-req-00) Sunil Iyengar A security I-D had been presented at IETF-63 (in Paris). The proposed L2 security method is based on an analysis of ULE. It had received many good comments at the previous meeting, but also some important questions from the IETF Security Area. This new I-D responds to these comments. It mainly describes threats and security requirements. The L2 security association is between an Encapsulation Gateway and the Receiver. A number of issues have been brought up on the mailing list. The authors are reading these and intend to work through them on the mailing list after the IETF meeting. The intention is to revise the document in December 2005, collect more feedback, and finally ask the WG to adopt it against the new Milestone. Marie-Jose Montpetit: The I-D speaks about MPEG-2/DVB and IP operators. What about content provider configuration? Martin Stiemerling: That's a good question. It should enter the picture somewhere. Marie-Jose Montpetit: All these people have configuration. Martin Stiemerling: If we talk about IP-level questions, then I am not sure they are involved, if this is at the service-level, then there is configuration. Gorry Fairhurst: Does anyone have NDP experience in very flat networks - i.e. scalability to large numbers of systems? Marie-Jose Montpetit: The Secure Neighbor Discovery WG (send) has work on scaling, captured in an RFC. This is important to the AR I-D. Martin Stiemerling: I will look into this point. Gorry Fairhurst: Is the I-D heading towards an Informational RFC. Should the outcome be part of another I-D? Martin Stiemerling: There can be recommendations on DHCP, NDP, arp, etc. Gorry Fairhurst: This seems to fall within the Chartered item on Address Resolution. We may finally need to consider whether this I-D should be combined with the AR draft. For now, I am very happy to see the I-D proceed, and would like to see contributions. 6. Presentation Security Method ------------------------------- (draft-cruickshank-ipdvb-sec-00) Sunil Iyengar This was the same I-D that had been presented at IETF-63. The authors will keep this draft alive while the requirements I-D matures, and update it. There will a new version -01 at the end of January 2007. 7. DVB-S.2 Framework -------------------- (WG Chair, Gorry Fairhurst) Gorry presented one slide showing the position of the new DVB-S.2 physical-layer within the protocol stack. This was expected to become a widely deployed standard and support existing ULE, MPE, etc via a transport stream interface. There was also a direct mapping of IP packets into the physical layer, known as the Generic Stream. This was under study in DVB-GBS, and was the topic for the next few presentations. There is an opportunity here to ensure that the two modes of DVB-S.2 have a common IP interface (e.g. based on ULE). 7.1 DVB-S2 Encapsulation GBS Activities ----------------------------------------- (prepared by Axel Jahn, presented by Gorry Fairhurst) Axel was responsible for S.2 IP Encapsulation work within DVB-GBS. The presentation described how DVB-GBS intends to define S.2 encapsulation requirements and evaluation criteria (see slides). The IP/S.2 group will send a copy of their document to the ipdvb mailing list. The group is asking for potential encapsulation methods. Next year, the comparison and evaluation would start. Many ipdvb WG members are already participating in this process, others are welcome via a DVB member. 7.2 Requirements for IP over DVB-S.2 ------------------------------------ (draft-cantillo-ipdvb-s2encaps-01) Juan Cantillo Juan presented a framework for IP encapsulation using Generic Streams. This version of the I-D has been completely rewritten, based on comments on and off the list. He presented a list of pending issues concerning the I-D. The next step is to propose this as a WG item. This is a hot topic, but does the WG scope cover DVB-S.2? Martin Stiemerling: What is the fragmentation problem? Juan Cantillo: ULE fragments SNDUs in MPEG-2, in S.2 BBframes have no defined method for flexible fragmentation. Martin Stiemerling: We already have ULE, is fragmentation out of scope? Gorry Fairhurst: I'm not sure this is the same problem, let's take questions about scope later. Josef Schmidbar: Are there any plans to utilize the different modulation options and link these to the IP QoS? Juan Cantillo: I am not aware of any work yet, most current systems also use MPEG-2 at constant rate. QoS could be added to the I-D. The authors would welcome ideas and text for this. Mark Watson: What is the relation between physical layer FEC and packet-level FEC? The physical layer is not necessarily the best place to expend effort to ensure overall reliability. Can the whole stack be optimised? Juan Cantillo: We are hopeful about driving ideas on cross-layer optimizations between L2 and L3. This topic needs to explored, and has many implications on IP. We are not yet talking about application optimisation, this is in the longer term. Gorry Fairhurst: You have to consider about applications/transport if you need to understand the loss/error requirements of specific flows. Packet FEC, is now an important IETF Transport Area topic, with documents in AVT and RMT WGs, and a BoF this IETF (FECFrame). This WG is not the only place in the IETF to address this topic. Juan Cantillo: There are potential some tools in S.2, that we can use for example the ISI field. Gorry Fairhurst (WG Chair): The Charter defines what the WG should do. Any change needs to be discussed with our Area Director (and the IESG). Anyway, the WG should not duplicate what is done in another organisation. Axel's contribution seems to clarify things. It seems to methat the I-D is more focussed on IP and there is no overlap. This may compliment work in DVB-GBS by bringing Internet experience. How many had read draft-cantillo-ipdvb-s2encaps-01? (6 people had read the document) How many people think the WG has expertise to work on this topic? (6 people had read the document) The authors should continue to revise this I-D, and the WG should pursue this topic on the mailing list. The protocol specification is not a valid WG item: It would therefore be premature to propose any Milestone in this space within the ipdvb WG. 8. ULE Implementation Status ---------------------------- (WG Chair, Gorry Fairhurst) Gorry presented a list of known implementations. Once ULE is published as an RFC, implementors should be encouraged to bring their code up to the proposed standard. There is another new implementor working on trial open source for a ULE Gateway. No details were available yet, but we should expect an announcement in a few months, this will complement the existing open-source Receiver. He asked if there were any news from other implementors? Josef Schmidbar: University of Salzburg had contributed to the Linux ULE release (now a standard feature of the kernel), and had recently discovered a small bug. The treatment of the D-bit (destination NPA/MAC address suppressed) was wrong when D=0 (specified MAC/NPA address) and a multicast L2 address was used. This could lead to unexpected packet loss for IPv4/IPv6 multicast. Gorry Fairhurst: Was this implementation going to be fixed? Josef Schmidbar: Yes, and it would be uploaded to the Linux distribution. 27 people were in the meeting room, and some also participated via jabber/audio. The session ended at 2:30pm.