Internet Video: Past, Present, and Future
LT6 (2/F), Yasumoto International Academic Park, The Chinese University of Hong Kong
Monday, August 12, 2013 (afternoon)
Presented by: Vyas Sekar (Stony Brook), Ion Stoica (UC Berkeley), Hui Zhang (CMU)
Tutorial ObjectivesThis tutorial will serve four main objectives catered to a broad audience spanning both industry and academia to address the following aspects of Internet video:
- Shedding light on the state of the art in Internet video delivery, and how the world has changed from traditional video/QoS research in networking.
- Provide a data-driven understanding of the current quality being delivered—-where are the bottlenecks, how does quality impact user experience, and steps that we can take to mitigate quality issues.
- Provide an overview of the tools and techniques that enable researchers and industry practitioners to bootstrap their entry into this space.
- Outline challenges and future directions for the next generation of Internet video research and deployment.
14:00-14:45 History and video ecosystem today
Perspective on video and QoS research: what did we get right, where did we go wrong?
Overview of content production, distribution, and player technologies being used today
14:45-15:30 How does quality impact engagement
Overview of industry-standard quality metrics
Brief overview of useful data analysis building blocks
Measurements showing impact of quality vs. engagement
Confounding factors that impact quality vs. engagement modeling (e.g., diversity in platforms, user quitting, regional biases etc)
15:30-16:00 Afternoon break + Informal breakouts
16:00-17:00 Deconstructing and mitigating video quality problems
State of art in video bitrate adaptation, challenges, and solutions
Longitudinal analysis of video delivery bottlenecks
Critical factors that impact video quality across genres, providers, locations
Multi-CDN optimization and potential for improvement
17:00-17:30 Concluding remarks and future research challenges
New trends and market forces; e.g., CDN/ISP federations
Future internet proposals and video (e.g., XIA, NDN)
Connections to SDN, cloud? Return of P2P?
Within the next few years, a significantly larger fraction of video consumption will be delivered over the Internet, with some analysis reporting that video will expand to become 91% of all Internet traffic by 2014. The new Internet video ecosystem is being shaped by the convergence of a few important technological and societal trends. First, as broadband speeds continue to increase, it drives video bit rates projected to increase from 500 Kbps-2 Mbps today to 10Mbps-20Mbps. Second, users and content providers now have easy access to the devices, software, and infrastructure required to create online HD-quality videos. In parallel, the emergence of popular online services, such as Hulu, Netflix, iTunes, and YouTube, has made huge libraries of multimedia content easy to share and disseminate. Finally, users employ a wide range of devices, such as smartphones, tablets, Internet-connected TVs, and set-top-boxes for viewing video.
The culmination of these trends means that we are in a position to design a easier-to-deploy and easier-to-evolve system than the 90s’ QoS/IP Multicast architecture which was designed under the assumption that the majority of traffic required network and server-level support for real-time QoS guarantees and IP multicast. Furthermore, the unprecedented scale of video adoption coupled with the ability to obtain fine-grained measurement insights from the client-side makes us uniquely positioned to leverage the power of “big data” techniques in network design.
We believe that the time is ripe for the networking community to revisit architectures for Internet video delivery. This tutorial highlight the unique characteristics of the video delivery ecosystem, provide key insights into the critical quality bottlenecks that users face today, and how providers can work around these quality problems will be timely.
About the Presenters
Vyas Sekar is an Assistant Professor at Stony Brook University. Previously, he was a research scientist at Intel Labs where he was a member of the Intel Science and Technology Center for Secure Computing. He received his Ph.D. from the Computer Science Department at Carnegie Mellon University in 2010 and his bachelor’s degree from the Indian Institute of Technology Madras, where he was awarded the President of India Gold Medal. His work has been recognized with best paper awards at ACM SIGCOMM and ACM Multimedia.
Ion Stoica is a Professor in the EECS Department at University of California at Berkeley, where he does research on cloud computing and networked computer systems. His past work includes the Dynamic Packet State (DPS), Chord DHT, Internet Indirection Infrastructure (i3), declarative networks, replay-debugging, and multi-layer tracing in distributed systems. His current research includes resource management and scheduling for data centers, cluster computing frameworks, and network architectures. He is the recipient of a SIGCOMM Test of Time Award (2011), the 2007 CoNEXT Rising Star Award, a Sloan Foundation Fellowship (2003), a PECASE Award (2002), and the 2001 ACM doctoral dissertation award. In 2006, he co-founded Conviva, a startup to commercialize technologies for large scale video distribution.
Hui Zhang is a Professor in the Computer Science Department at the School of Computer Science at Carnegie Mellon University. He won the National Science Foundation CAREER Award in 1996 and the Alfred Sloan Fellowship in 2000. He was elected a fellow of ACM in 2006. During 2000 - 2003, Prof. Zhang was the Chief Technical Officer of Turin Networks (merged with Force 10 Networks in 2009 and acquired by Dell in 2011). He is also the co-Founder and CEO of Conviva Inc.