Emerging Trends in Live Streaming: Challenges and Opportunities
The rise of live streaming is transforming the digital landscape, impacting network resources and audience engagement. With increasing demand for high-quality live content, network operators face challenges in scalability and capacity. The transition to live streaming reflects shifting consumer behaviors and the need for authentic, real-time experiences. The COVID-19 pandemic has further accelerated this trend, driving innovations in virtual gatherings and e-commerce. As the industry continues to evolve, network operators must adapt to meet the demands of mass audience live streaming.
Download Presentation
Please find below an Image/Link to download the presentation.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. Download presentation by click this link. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.
E N D
Presentation Transcript
TreeDN: Tree-based CDN Architecture for Mass Audience Live Streaming Lenny Giuliano <lenny@juniper.net> Non-Juniper
Are we approaching an inflection point for network resources consumed by live streaming? Will we ever? If yes to either, what can network operators do about it? Non-Juniper
Live Streaming: By the Numbers First, some simple facts: HD Bitrate: 4-8 Mbps 4K Bitrate: 35-45 Mbps 8K Bitrate: 80-100 Mbps VR Bitrate: 500Mbps-5.2 Gbps!!! Akamai Traffic Record (Apr 2020): 167 Tbps Akamai estimated to deliver 15-30% of total web traffic 167Tbps / 40Mbps = 4.2m simultaneous 4k viewers to set new traffic record 3 Non-Juniper
* Source: Jake Holland (Akamai) @ NANOG 79 Non-Juniper
Can Brute Force Unicast (BFU) Keep Up?? TV Audience of average NFL game: ~15M 15M viewers @ 40Mbps = 600Tbps!! Latest NFL TV deal includes games exclusively avail through streaming TV Audience of other major events: Super Bowl 100m 2015 Cricket World Cup (India played): 200-300m 2018 FIFA World Cup Finals: 500m Cord Cutting Evolution- live, linear TV is the last frontier YouTubeTV, SlingTV, HuluTV, etc Live, linear TV is not dead yet! 5 Non-Juniper
Whats old is new: Live streaming is Trending Brand Transparency and Authenticity: due to spontaneous and un-editable nature, live streams are perceived as more authentic and drive greater emotional engagement than on-demand Viewers spend 8X longerwith live video than on-demand, 67% of live viewers are more likely to make a purchase and best ROI from live video compared to any other social media platform The COVID Pandemic has pushed the need for solutions that enable people to gather virtually (eLearning, fitness/wellness, conferences, worship services, sporting events, drinks with friends, etc) Between January and August 2020, the number of live-streamed events hosted increased by 1468% Live stream watch time has increased by 250% Live Streaming E-Commerce Is The Rage In China: $129B by end of 2020 (from $29B in 2018) 6 Non-Juniper
Conclusion: Live linear video is here to stay but, IP (unicast) is not well suited to multi-destination traffic IP Multicast was created 2+ decades ago to solve this problem So, what went wrong? 7 Non-Juniper
In the beginning, there was unicast $$$$!!! Non-Juniper
then came Multicast Non-Juniper
But, there was a problem with multicast Internet Unicast-only Mcast-enabled Mcast-enabled Unicast-only Mcast-enabled Unicast-only Unicast-only Unicast-only The All or Nothing problem Non-Juniper
What to do? CDNs! Content Delivery Networks became popular Almost as good as multicast, and it just works! Non-Juniper
Comparison: Unicast, CDN and Multicast Cost (audience) for the Content Provider CDNs don t eliminate the BFU problem, they just distribute it EUMETSAT: multicast is an order of mag cheaper than unicast CDNs!! Unicast Cost CDN Multicast Audience Non-Juniper
Internet Mcast and IPv6: Technological Cousins Any argument for/against IPv6 applies just as well to Internet Mcast It s not needed: NAT, CDN It is needed: v4 address exhaustion, inefficiency of duplicated video streams Both suffer the all or nothing problem Both require new protocols/extensions of old ones Both born in the 90s, struggled through adolescence, un/underemployed as young adults Both require faith in the unseen Both do accomplish little on their own, but enable big things Non-Juniper
So Whats New? Internet Multicast has been around since the 90 s and hasn t gained much traction, so why will this time be different? Biggest problems of Internet Multicast: 1. All or Nothing problem 2. Multicast is too complex problem 3. Content-Audience Chicken and Egg problem Non-Juniper
Heres Whats New! ASM Deprecation (RFC8815/BCP229)- SSM-only, eliminates most of the complexity of multicast BCP already applied in I2, solves the Multicast is too complex problem (#2) Overlay Networking with AMT Addresses fundamental All or Nothing problem of multicast (#1) Tunneling is fashionable- Overlay Networking is cool now! Also solves the Chicken and Egg problem of multicast (#3) All the pieces are in place for a working solution TODAY! Non-Juniper
AMT- How it works Gateway (client) uses Anycast to discover nearest relay. UDP-encapped IGMP report is sent to relay at the edge of the mcast network. Content Owner Mcast Enabled ISP Unicast-Only Network Relay GW Mcast Traffic Mcast Join UDP-IGMP Request Mcast Enabled Local Provider Non-Juniper
AMT Unicast Edge Network Content Owner Mcast Enabled ISP Mcast join is sent by relay on behalf of the unicast receiver. Unicast-Only Network Relay GW Mcast Traffic Mcast Join UDP-IGMP Request Mcast Enabled Local Provider Non-Juniper
AMT Unicast Edge Network Content Owner Content Owner Mcast Enabled ISP Mcast Enabled ISP Mcast stream is replicated and encapped in UDP to the unicast receiver. Unicast-Only Network Unicast-Only Network Relay GW Mcast Traffic Mcast Join UDP-IGMP Request UDP Ucast Stream Mcast Enabled Local Provider Mcast Enabled Local Provider Non-Juniper
AMT Unicast Edge Network Buy a bigger pipe, or enable multicast, or both? (AMT is a stopover, not a destination) Content Owner Content Owner Mcast Enabled ISP Mcast Enabled ISP Unicast-Only Network Unicast-Only Network Relay GW GW GW Mcast Traffic Mcast Join UDP-IGMP Request UDP Ucast Stream Mcast Enabled Local Provider Mcast Enabled Local Provider Non-Juniper
CDNs without Multicast Non-Juniper
CDNs with Multicast: TreeDNs Replication point closer to receivers, only to relays with nearby receivers If deployed on existing network infra (CDN-on-a-Chip): $0 capex and maybe $0 opex, too AMT Relay AMT Relay AMT Relay AMT Relay AMT Relay Non-Juniper
Steps for World Domination TreeDN in 5 steps: Step 1: AMT Relays Step 2: AMT GW implementations Step 3: Portal for mContent Step 4: Off-Net Sourcing Step 5: ??? Profit!! 22 Non-Juniper
Step 1: AMT Relay Deployment Step 1: AMT Relay Deployment 4 public AMT relays deployed on the MBONE today Thomas Jefferson HS, George Washington Univ (2), Univ of Kansas Would love to have more relays! Had several I2/GEANT institutions express interest in deploying relays Relay discovery done with DNS (amt-relay.m2icast.net) Relay Resilience: done by application layer- VLC picks one, and if no packets arrive, picks the next one in the list Relay Optimality: TBD Hint: will probably use DNS 23 Non-Juniper
Step 2: AMT Gateway Implementation Goal- build a modern AMT GW implementation that s easiest enough for any Internet user to use Short term: AMT GW patch available in VLC 4.0.0 today Download and use it now: https://nightlies.videolan.org Long term: MTTB Akamai s work @ IETF and W3C 24 Non-Juniper
Step 3: Multicast Content Portal https://multicastmenu.herokuapp.com Webpage/Portal with a list of active multicast streams Multicast Menu by Lauren Delwiche/TJHS Built on William Zhang s MBONE curation script Script periodically crawls through all I2 and GEANT routers at Looking Glass for multicast routes Dynamically populate the list of available streams on the MBONE on a webpage, launch VLC to open Can manually add streams as well 25 Non-Juniper
Step 4: Off-Net Sourcing Sourcing from MBone is available today (On-Net) but most sources will be on unicast-only networks (Off-Net) Off-Net sourcing- users on unicast-only nets can stream to a translator on MBONE, which converts stream to multicast Then accessible natively to on-net receivers and via AMT to off-net receivers Demonstrated successful off-net sourcing with Unicast2Multicast Translator @ IETF112 26 Non-Juniper
TreeDN Architecture: Putting it all together Internet Unicast MBONE (GEANT, I2, etc) Mcast Source Off-net Source U->M Translator Native Receiver Relay Relay AMT Relay Off-net Receiver Non-Juniper
Whats in it for Network Operators?? What s in it for Network Operators?? Combination of exploding live audience sizes plus increasing bitrate streams suggest it may be time for a new approach for CDNs Creates viable business model for live streaming to mass audiences Market for mass streaming could be huge Can offer new livestream replication services at a fraction the cost of unicast-only CDNs Built into existing router infra, no new servers/blades to rack/stack/power/connect ($0 capex/opex) Democratizes and decentralizes content sourcing Is it healthy for the Internet (and society) that a small handful of companies control nearly all content distribution? 28 Non-Juniper
References References Unicast2Multicast Translator https://github.com/JNPRAutomate/unicast2multicast-translator Demo from IETF112 Lauren s Multicast Menu https://multicastmenu.herokuapp.com Source code: https://github.com/Laur04/multicast-menu Natalie/Wayne s VLC with AMT GW build (and usage notes) https://github.com/Juniper/amt-vlc William s AMT Relay Deployment docs https://github.com/willzhang05/senior-research MTTG Slack group- send email to lenny@juniper.net to join Non-Juniper
