Early experience with an internet broadcast system based on overlay multicast
read more
Citations
CoolStreaming/DONet: a data-driven overlay network for peer-to-peer live media streaming
CoolStreaming/DONet: A Data-Driven Overlay Network for Efficient Live Media Streaming
DONet: A Data-Driven Overlay Network For Efficient Live Media Streaming
Opportunities and Challenges of Peer-to-Peer Internet Video Broadcast
Characterizing residential broadband networks
References
RTP: A Transport Protocol for Real-Time Applications
A case for end system multicast
Measurement study of peer-to-peer file sharing systems
Scribe: a large-scale and decentralized application-level multicast infrastructure
Scalable application layer multicast
Related Papers (5)
Frequently Asked Questions (13)
Q2. What are the primary goals in supporting NATs and firewalls?
The primary goals in supporting NATs and firewalls are: (i) enable connectivity, a generic problem shared by many applications wishing to support these hosts and (ii) address protocol-specific enhancements to become “NAT/firewall-aware” to improve efficiency and performance.
Q3. How many users see loss for less than 10% of the session in high quality video?
Over 70% of the users see loss for less than 10% of the session in low quality video, but only 50% of users see loss for less than 10% of the session for high quality video.
Q4. What are the main issues that are not considered in protocol design studies?
The authors address a wide range of issues such as support for heterogeneous receivers, and NATs and firewalls, which are not typically considered in protocol design studies.
Q5. What are the two key factors that capture the quality of resources in an environment?
Two key factors capture the quality of resources in an environment: (i) outgoing bandwidth of hosts, which directly bounds the number of children hosts can take; and (ii) the presence of NATs and firewalls which places connectivity restrictions on parent-child relationships.
Q6. How many hosts see loss for less than 5% of the session in the audio and low quality?
over 90% of the hosts see loss for less than 5% of the session in the audio and low quality streams, and over 80% of the hosts see loss for less than 5% of the session in the high quality stream.
Q7. How do the authors evaluate the potential of contributor-aware policies?
The authors evaluate the potential of these policies by conducting a trace-based simulation using the group membership dynamics pattern from the Slashdot broadcast.
Q8. How does the probe determine the amount of bandwidth resources each host can contribute to the overlay?
In order to assess the amount of upstream bandwidth resources each host can contribute to the overlay, the authors ask the user to choose whether or not it has at least a 10 Mbps up-link to the Internet.
Q9. How much of the quality index is a result of NATs and firewalls?
In light of their experience, NATs and firewalls can constitute an overwhelming fraction of a broadcast (for example, 50%-70% in Slashdot ), and thus significantly lower the Quality Index .
Q10. What is the heuristic for dynamically adjusting the detection time?
The authors have implemented a simple heuristic for dynamically adjusting the detection time, involving an increase if several parent changes have been made recently, and a decrease if it has been a long time since the last parent change.
Q11. How many viewers were turned down for the lack of support for NATs and firewalls?
Their initial prototype did not include support for NATs and firewalls, and the authors were motivated to address this as the authors consistently needed to turn down over 20 − 30% of viewers in their early broadcasts for the lack of such support.
Q12. What is the appropriate coding scheme for a video stream?
Internet hosts are highly heterogeneous in their receiving bandwidth, thus a single-rate video coding scheme is not the most appropriate.
Q13. How does the player hide the fact that the parent of the overlay proxy changes over time?
To hide from the media player the fact that the overlay parent changes over time, the authors direct the player to a fixed localhost:port URL which points to the overlay proxy running at the same host.