**Session Date/Time:** 21 Jul 2026 07:00 # [MOPS](../wg/mops.html) ## Summary The Media Operations (MOPS) Working Group met at IETF 126. The session featured an interactive industry panel discussing local geography and operational experiences with video streaming, an update on the active working group draft `draft-ietf-mops-network-overlay-impacts`, an industry update on the Streaming Video Technology Alliance (SVTA), and a discussion regarding a proposed charter update to transition the working group from experimental to permanent status. --- ## Key Discussion Points ### 1. Administration & Chair Updates * **Chairs:** Leslie Daigle, Jana Iyengar * **Notetaker:** Chris Lemmons * **Slides:** [MOPS Chair slides](https://datatracker.ietf.org/meeting/126/materials/slides-126-mops-mops-chair-slides-00) * Leslie Daigle opened the meeting, presenting the Note Well guidelines and checking for agenda bashes. No modifications were made to the agenda. --- ### 2. Industry Panel: Local Geography and Operational Experiences Jana Iyengar moderated a panel featuring Matt Green (Sky), Guillaume Bichaud (Broadpeak), and Steve (BT) focusing on the operational realities of delivering streaming video and game downloads at scale. #### Traffic Peaks and Network Scaling * **Matt Green** highlighted that football (soccer) matches (e.g., Premier League, Champions League, World Cup) drive the largest streaming peaks on Sky's network. Large game downloads (e.g., *Fortnite*, *Call of Duty*) also cause massive traffic spikes, and can be particularly challenging when they coincide with live sporting events. While CDN caches are deployed deep to offload core networks, the proliferation of independent CDNs run by individual streaming services leads to redundant infrastructure. * **Guillaume Bichaud** shared telemetry from Broadpeak deployments during the recent World Cup, showcasing successful scaling via Multicast Adaptive Bitrate (MABR). In South America, peak concurrent streams surpassed 1.6 million sessions (over 1.5 to 2 Tbps of multicast traffic) with 98% efficiency (i.e., only 2% fallback to unicast). However, he noted a major bottleneck occurs right after matches when users simultaneously switch channels, creating massive spikes on CMS platforms and load balancers. * **Steve** observed that while game downloads can tolerate slight delays, video streaming is highly sensitive to congestion. BT is closely tracking the UK regulator Ofcom’s green paper suggesting a complete switch-off of terrestrial broadcast by 2034. Transitioning all live and emergency broadcasts to IP networks will drastically shift capacity planning. BT is trialing a variant of MABR designed to integrate over-the-top (OTT) providers and CDNs. * **Last-Mile vs. Core Bottlenecks (Q&A/Chat):** * Alan Frindell asked where unicast delivery hurts most. Matt Green and Steve explained that the highest incremental cost is at the first aggregation point just before the access network (where access switches handle backhaul). * Chat discussions between Will Law, Mike Blanche, Kyle Rose, and Leslie Daigle highlighted that while Europe enjoys high Fiber-to-the-Home (FTTH) coverage (near 80%), other regions like the US (around 30% FTTH) still deal with congested shared loops (cable, LTE, satellite) over the last mile. #### Advertising & Ad Insertion * **Steve** explained that while BT doesn't control ads directly, they experiment with preloading personalized ads to home gateways over unicast or multicast ahead of ad breaks to avoid sudden unicast peaks. * **Matt Green** noted Sky uses client-side dynamic ad insertion via manifest manipulation. Integrating personalization with multicast remains a challenge because dropping back to unicast for personalized ads collapses the traffic spike into a narrow time window, retaining high peak capacity requirements. * **Guillaume Bichaud** stated Broadpeak heavily utilizes Server-Side Ad Insertion (SSAI) and Ad Insertion on the Edge (AGAI). For massive sporting events, personalization is minimal because high ad prices favor broad national ads, allowing multicast to operate efficiently without generating massive unicast spikes. #### Future Evolution & Standardization * **Guillaume Bichaud** identified Media over QUIC (MoQ) as a highly promising future architecture for efficient live streaming at scale due to its standardized framework and potential for low-latency delivery. * **Mike Blanche** raised a question on how to streamline peak forecasting and scheduling to improve interconnection between content providers and access networks (referencing upcoming EU Digital Networks Act regulations). Matt Green supported adopting an API-based interface to replace manual, unscalable spreadsheet-sharing. Guillaume Bichaud mentioned that SVTA’s Open Caching APIs are designed to address this, introducing "capacity-based" billing models to better handle peaks. --- ### 3. Working Group Draft: `draft-ietf-mops-network-overlay-impacts` * **Presenter:** Sanjay Mishra * **Slides:** [Network Overlays update IETF126](https://datatracker.ietf.org/meeting/126/materials/slides-126-mops-network-overlays-update-ietf126-00) Sanjay Mishra presented the updates made to version -04 of the active working group draft `draft-ietf-mops-network-overlay-impacts` (co-authored with Glenn Deen). The revision addressed comprehensive feedback received from David Schinazi: * **Introduction Rework:** The section was tightened to focus clearly on the document's scope, the tension between user privacy enhancements and operational blind spots, and explicitly clarifying that this is a problem statement document (mitigation strategies are out of scope). * **VPN Clarification:** The explanation of VPNs was moved from the appendix to Section 2.2. The draft now clarifies that while VPNs behave as overlays, the core operational distinction is transparency and connection autonomy rather than the protocol type. * **Typos:** A flipped digit in RFC 7624 was corrected. * **Scope Discussion:** Tim Chown asked if the draft should address VPN usage by minors to bypass social media/streaming bans. Glenn Deen clarified that access bypass is a regulatory and policy issue, which is out of scope for this document's focus on operational network conflicts. * **Next Steps:** The draft has no blocking issues. The authors will issue one final revision (-05) to eliminate minor text redundancies and integrate an open PR from Jay Robertson. --- ### 4. SVTA MOPS Update * **Presenter:** Glenn Deen * **Slides:** [SVTA MOPS update IETF126](https://datatracker.ietf.org/meeting/126/materials/slides-126-mops-svta-mops-update-ietf126-00) Glenn Deen provided an overview of the latest developments in the Streaming Video Technology Alliance (SVTA): * **Open Caching:** Work continues on defining operational extensions for IETF CDNI specifications. * **Edge Compute / QoE:** Projects are underway to standardize edge cache hosting and analytics. * **QUIC vs. TCP Report:** The Networking and Transport group published a long-running Proof of Concept (PoC) report indicating that QUIC (HTTP/3) performance is comparable to HTTP/2 ("no worse, no better") for traditional video streaming. * **AI Workshop:** SVTA will host an engineer-to-engineer workshop and hackathon on "AI in Media Distribution" in Rennes, France, in October 2024. --- ### 5. Charter Review & Milestones * **Presenter:** Leslie Daigle * **Slides:** [MOPS Chair slides](https://datatracker.ietf.org/meeting/126/materials/slides-126-mops-mops-chair-slides-00) Leslie Daigle introduced a proposal to update the MOPS charter. * **Proposed Change:** After seven years of successful operation and publishing three RFCs, the chairs propose removing the historical sentence: *"The IESG is establishing this working group on an experimental basis..."* to transition MOPS to a permanent operational working group. * **AD Feedback:** Éric Vyncke (Internet AD) strongly supported the change, noting that the MOPS model has been highly successful and was used as a blueprint for other operational groups like IoT OPS and SRv6 OPS. * **Milestones:** Tim Chown suggested adding new milestones, noting the last archived milestone was from 2022. Éric Vyncke responded that while the core charter text requires a formal IETF last call, milestones are indicative and can be updated flexibly by the chairs and AD. --- ## Decisions and Action Items * **`draft-ietf-mops-network-overlay-impacts`:** * **Decision:** The working group agreed that the draft is technically complete and ready for Working Group Last Call (WGLC) once editorial cleanups are finished. * **Action Item:** Sanjay Mishra and Glenn Deen to publish version -05 addressing minor redundancies and incorporating Jay Robertson's PR. * **Action Item:** Chairs to initiate WGLC once version -05 is published. * **Charter Update:** * **Decision:** The working group agreed to proceed with updating the charter to remove the "experimental" designation. * **Action Item:** Chairs to post the proposed charter text to the MOPS mailing list for a two-week review period before submitting it to the IESG. --- ## Next Steps 1. **Mailing List Discussion:** Initiate review of the updated MOPS charter text on the mailing list. 2. **Draft Publication:** Publish version -05 of `draft-ietf-mops-network-overlay-impacts` and start WGLC.