Measurement and Analysis for Protocols Research Group (maprg) Minutes for IETF-97 (Seoul) Thursday, Nov 17, 9:30-11:00 (Morning session I) Room: Grand Ballroom 2 Chairs: Mirja Kühlewind and Dave Plonka Rethinking Broadband Performance using Big Data from M-Lab - Xiaohong Deng ------------------------- (short talk, no comments or questions) Video at the Edge: A Measurement Study - Kathleen Nichols (remote) ------------------------- (Comment on Netflix Video Delay Variation, from jabber room, since Kathleen uses TSval for clocking and would like higher frequency): Praveen Balasubramanian: note that boosting the clock frequency to 1ms has power implications. Martin ???: Is the apple vs chromecast thing an app or a stack issue? Kathleen: At least a little is the app, different numbers of streams. Flow startup is different, too, so that's Tommy Pauly: Some behavior, we send video streaming through shared system libraries. On google, they have control. Jana: Wish we had more time. Kathleen: I'll try to write more of this up Jana: Good TS allows more extraction. How much more fidelity do you get from TSval than simple acks? Kathleen: No OWD without TSval. Brian: What's realtime? Kathleen: Can do this with thirty seconds delay, but these are really thirty second batch windows. Could do with less, estimator can determine it has enough data. Aaron: Please come to Chicago. Traffic Policing in the Internet - Yuchung Cheng and Neal Cardwell ------------------------- Tommy Pauly: I doubt policers would do this. Maybe they could use ECN to signal this. Niel: We have. In the long term, worth looking into. Praveen (via jabber): Is there delay buildup before drops on policing? Neal: No. We would prefer they shape it. H2 performance analysis in cellular networks - Moritz Steiner (remote) ------------------------- Yuchung: BBR fixes this. Will forward you data. Mirja: Please come to Chicago and bring data.