SAMRG (Scalable Adaptive Multicast Research Group) IETF-81 Chairs: John Buford and Thomas Schmidt Note-taker: Matthias Waehlisch Thomas: Presented the agenda. A Common API for Transparent Hybrid Multicast Matthias presented draft-irtf-samrg-common-api-02 Matthias: First reviews have been performed, first in private, but we have asked to forward this to the list. Thomas: Folks, we will need more reviews of this mature draft. John: Will you consider this to be forwarded to be published as an RFC? Matthias: Yes, if reviews have been completed and positive. John: May be we can have an RGLC before Taipei Implementation report Matthias presented Release of the System-centric Middleware Component for Universal Multicast Matthias: Please download, try out and provide feedback/contribute. How are first experiences? ZhigangSun: We downloaded and installed, want to add technology modules. We still have questions on the documentation. Matthias: Code should be fully documented, but please send any question. We will provide all additional information needed. John: Why do you have packet loss for the IP path but not for the Overlay? Matthias: IP handles many more packets, probably it's a small buffer overflow, we have to check the details. John: Would be good to add a section on implementation experiences in your API draft. Matthias: Yes, we can put this to the Appendix. Application Layer Multicast Extensions to RELOAD. John: Presented draft-kolberg-sam-baseline-protocol. Lars: Regarding the IANA considerations, if you allocate RELOAD code points you extend RELOAD. You cannot do this. I suggest to send an email to RELOAD authors to include code points for private usage. Marc Petit-Huguenin: Last call ended this week. Thomas: Can you apply different overlay multicast mechanisms? I wonder whether several functions you describe are sufficiently generic to work for most ALM schemes? John: We picked Scribe as a first point because we think this a feasible mapping. But others should follow. Problem Statement: Streaming Media Delivery. Tao: Presented draft-litao-p2mpsmd-problem-statement-01. Thomas: What does Delay Factor mean? Tao: It is defined in RFC 4445. Matthias: Check back with PPSP problem statement draft. Lars: Regarding RFC 4445, look at the IESG disclaimer. There are severe doubts regarding the proposed metrics. It's not a measure, the IETF would use. You can also check with the comments in the RFC tracker. Thomas: Have you compared with well-known mechanisms (e.g., IP layer multicast, overlay multicast). Tao: ... For Labelcast we have not such comparisons. ... Labelcast Protocol. Zhigang Su: Presented draft-sunzhigang-sam-labelcast-02. Matthias: What do you mean with group member distribution? What is the difference compared to network topology? Zhigang: The group member distribution means the multicast distribution tree. The network topology is below. Lars: How do you want to use the protocol in NAT and firewall scenarios? Zhigang: UDP checksum field are actually not used Thomas: What Lars means is that you are replacing the transport protocol. And this is something that is considered by firewalls. Thomas: Maybe you consider the wrong layer when replacing the transport protocol. What you actually aim at is a label mapping + switching. I suggest to use the IPv6 flow label for it. With out of band control plane, you can simply map the label to your forwarding instructions. The only lack I see are your in-line time measurements. Maybe you can also use the flow label for measurements in a clever way? This would also save you from replacing the transport. Lars: Time stamping every packets is very expensive. You'll make the routing slow. Zhigang: It produces no latency. Lars: You make label switching slower, because you add additional burdens. Discussion on future research directions Thomas presented topic list as outcome of Prague discussion and several side contacts Ronald: The charter also mentioned MANETs. Is this of interest for the working group. John: At the moment nobody is working on MANETs in this working group, but the topic is still on the agenda. Ronald: We have worked on SMF, Simple Multicast Forwarding for MANETs. Thomas: The integration aspect could possibly of interest to SAM, i.e., a joint multicast routing in the fixed Internet and a MANET.