IETF 107/Virtual Interim Tues, Apr 21, 2020 9-11AM PDT Meeting held jointly with PIM WG (PIM went first) Recording: https://youtu.be/5NwkTuWGs-M?t=5231 Note taker: Olufemi Komolafe ----- Notes take in etherpad at https://etherpad.ietf.org:9009/p/notes-ietf-107-pim Text (from 4/27/20) pasted below for reference: MBONED agenda: Status of WG items draft-ietf-mboned-ieee802-mcast-problems-11 Mike McBride: Got some good input from Linus Lussing draft-ietf-mboned-dc-deploy-09 Mike McBride: Keen to have more feedback regarding dc-deploy-draft. Happy to work on Jake's feedbac or let draft go if that's the WG's concensus Lenny Giuliano: if enough interest, authors to rev based on latest comments; if not enough interest, draft can fade off intto the sunset Jake Holland: Draft of interest and within scope but could be improved Warren Kumari: Can we recruit some volunteers to review the draft. Volunteers => Greg Shepherd, Dino Farinacci Chairs, 5 min Multicast to the Browser: draft-ietf-mboned-ambi-00, draft-ietf-mboned-cbacc-00, draft-ietf-mboned-dorms-00 Jake Holland, 30 min Dino Farinacci: Use asymmetric keys? But no good algorithms that are fast enough Jake Holland: Performance of security suites prohibitive. Probably worthwhile writing up that approach for when fast enough security approaches become available Dino Farinacci: Different mechanism to do authentication and authorization since authorization can be done much quicker? Jake Holland: Authentication is crucial Toerless Eckert: Isn't AMBI really necessary for only DRM-encumbered receivers? Jake Holland: There are some other scenarios when AMBI is needed Toerless Eckert: Broadcast DRM traffic well-resourced and deployed. Isn't that the answer to that threat model? Jake Holland: Diffcult to check in code to browser relying on only DRM Greg Shepherd: Perhaps consider taking to Nanog or other forums with carriers Jake Holland: Seeking carriers/content owners to particpate in POCs and trials Multicast to the Grandma (MTTG) Update Lenny Giuliano, 25 min Greg Shepherd: "People don't hate multicast they hate PIM", i.e. the problem is not network replication but rather the complexity designed to facilitate the replication Lenny Giulano: "People don't hate PIM SSM they hate PIM ASM" Toerless Eckert: PIM-SSM is simple enough for all broadcast. BIER is a great simplification over PIM-SM and enables new "sender-control-applications". Multicast via BGP is for operators that never learned a second language beside BGP ;-) Jake Holland: PIM SSM easier to follow that MBGP VPN. The BIER drafts are interesting and solving per-hop scale issues and reducing operational complexity is worthwhile. Toerless Eckert: Complementary to Jake Holland's effort. How does it interact with Verizon's presentation where the use of a local agent is used to change multicast traffic to unicast so browsers are unaltered Gyan Mishra: Rather than developing in-house from scratch, consider other approaches (e.g. webRTC) but ultimately approach was to use Ramp product Jake Holland: Key to be able to deploy multicast to the actually browser itself Greg Sheperd: Requirement that source be multicast is not ideal Easier to have concept of multicast for distribution but no reason why sources can unicast traffic to the FHR Dino Farinacci: Does not really matter how data is sourced, as long as multicast service model used for distribution. Can be flexible, depending on constraints of source applications. Also need to deal with NAT traversal if source is smartphone which is harder for multicast source than unicast Dino Farinacci: Need someone to write app to take video traffic from a phone and send to the network, in either unicast or multicast Leonard Giulano: Looking for potential open source apps