Transcript Tracker New Peer Peer Peer

Problem Statement of Peer to
Peer Streaming Protocol (PPSP)
IETF76@Hiroshima
Draft-zhang-ppsp-problem-statement-05
Yunfei Zhang
Ning Zong
Gonzalo Camarillo
James Seng
Richard Yang
Richard Alimi
Hirold Liu
Alberto J. Gonzalez
Agenda
•
•
•
•
•
•
•
Why Streaming?
Why P2P streaming?
Why Standards?
PPSP Scope
Open efforts related to PPSP
Co-efforts in other standard bodies
Conclusion
Why Streaming?
•
Source：Technology Trends,
June 2008,Morgan Stanley
Research
Morgan Stanley View:
Video streaming is the main
contributor for global IP traffic growth
in the following years, accounting for
50% of the total traffic.
The Compound Annual Growth Rates of Internet video streaming to
PC and TV are 56% and 97% .
Why P2P Streaming
• Scalability
• Each consumer is also a supplier
• Robustness
P2P Streaming Success
• Real-world examples
– PPLive
• 110m users, 600+ channels
• 2 million concurrent peers/6 million altogether in China’s 60
Anniversary National Day live broadcasting
• 20%-30% outside of China (10-15% in US), >200 countries
– PPstream
• 70m users, 340+ channels
• 6 million concurrent peers/10 million altogether in China’s 60
Anniversary National Day live broadcasting
– UUSee
• 4-5 million concurrent online peers during Olympic Games
• 2~3 million concurrent online peers in China’s 60 Anniversary
National Day live broadcasting
– CNN (OctoShape)
• Obama inauguration ~300K concurrent peers by OctoShape
Why Standards in P2P Streaming?
Technical feasibility: strong similarity among
major systems
• Tracker-based architecture
• Similar tracker and peer communication
process, and inter-peer communication
process
(See survey and measurement draft)
Standards => focusing on key issues, not
reinventing the wheel
Why Standards in P2P Streaming?
User desire:
“… broadcasters from the BBC to Germany’s ARD just seem to love
the idea of ditching their proprietary platforms.”
-- Johan Pouwelse, scientific director of P2P Next
“UUSee will start to build an open platform and would like to
participate in open protocols to cooperate with content providers,
operators and many more participants for a better p2p streaming
service.”
-- Zhu Li, CEO of UUSee
“The biggest feature of Mobile Market is its openness. We welcome
more participants involved in the Mobile market development with
standards.”
---Jianzhou Wang, President of China Mobile
How Standards Helps:
Use Case (1)
• P2P streaming
vendors often use
CDN to alleviate peak
pressure and achieve
better QoS
• Target: CDN nodes
can act as Superpeers, which interact
with trackers and
peers seamlessly
– Integration with
multiple CDNs is easy
with standard protocols
How Standards Helps?
Use Case (2)
• Network operators deploying open and
competitive Internet streaming service
– Operator can run its own streaming service or
cooperate with P2P streaming vendors
– User convenience: unified peer/client software
• In particular, for mobile devices
– Service can be deployed based on 3rd party components
(e.g., CDN offerings)
Problem
• Lack of open and standard protocols
• Open set of P2P streaming protocols is
needed: PPSP
Abstracts of Current Practice
Tracker
New Peer
peer request
peer report
Peer
buffer map exchange
peer report
Peer
buffer map exchange
Peer<->Tracker
Peer<->Peer
referring to survey slides
Features of P2P Streaming Systems
• Scale to a magnitude of up to millions of
nodes
• Real time
– Timely and continuous streaming delivery
– Limited start-up delay and transmission delay
– Handles demanding scenarios
– 1.5Mbps for TV quality
– 400kbps in current Internet
P2P Streaming Layer Architecture Overview
Application Layer
Play-out Layer
Start
Pause
Stop
Information Layer
Report
Publish
Statistics
Communication Layer
Bootstrap
Transport Layer
Neighbor Communication
Tracker Communication
PPSP Scope
• Develop new protocols; existing standards (e.g., HTTP) will be reused as appropriate
• Starting Point: Communication layer
– Tracker Communication
• peers registration to the system
• peer list request from or peer information report to the tracker/
– Neighbor Communication
• peers exchange data availability and gossip for more peers
• chunk description for transferring data amongst peers
• Secondary Specifications: Information Layer
• Report
– peers report information to the tracker.
– peer inbound and outbound traffic, amount of neighbor peers,
video quality parameters.
• Publish
– publish contents to streaming provider
– content description, type, creation time, source location,
advertisement policies, etc..
Out of Scope
• Play-out layer (e.g. Start, Stop, Pause)
• Defining new data transport protocol
Open efforts related to PPSP(1)
• UUSee:
– One of the largest p2p streaming vendors worldwide
– Just several days ago, its CEO expressed intention to
make an open p2p streaming platform in 2009 China
Internet conference
– Meet with its CTO, Hirold Liu about the PPSP work.
Liu stated that UUSee was looking for the standards
of P2P streaming, esp. in mobile environments
– Different peer capacity evaluation
• Different network environment description
• New peer information message
Open efforts related to PPSP(2)
•
•
•
•
•
CoolRuc: Audiovisual Live Streaming based on P2P
Project:TRILOGY: fuTuRe Internet technoLOGY in Europe
Prierd:2007 – 2009
Participants: i2CAT, UPC, UPF, URL
Aim: Standardization
PPSP & CoolRuc
• CoolRuc Protocol standarization points:
– Rendezvous communication (signaling)
• Retrieval of channels which are resources to be published
and discovered over the network
– Tracker Communication (signaling)  Tracker protocol
• Retrieval of initial list of peers
– Peers communication (signaling)  Peer Protocol
• Exchange of Buffer Map
• Chunk exchange (transference)
– Statistics
• Log messages
• More requirements are added in the requirement draft
Open efforts related to PPSP(3)
• Goalbit-Open source tracker-based p2p streaming system
• Developed in 2009
• Goalbit Transport Protocol is well suit with PPSP scope(Paper published
in Sep.2009)
• Peer and tracker communication
• Peer and Peer communication
• “It would be great for us, to help in defining a standardized PPSP.”--Pablo Rodríguez-Bocca,Chief of Goalbit Project
Open efforts related to PPSP(4)
• China Mobile-DSN,
Distributed Services Network
– One part of DSN is for open p2p
streaming service platform and
protocols
– Participants (Streaming): China
Mobile, ForceTech,Huawei
– DSN is both an open project and
an ongoing ITU standard for
NGN
Co-efforts in other standard bodies
• ITU:Q:DSN in SG13
– Period:2009~2012
– Scenarios, Requirements and Architecture of DSN
– Related to PPSP: The streaming scenarios, requirements and
architecture
• 3GPP: IMS based P2P CDS in SA1
– Period: 2009~2010
– Study p2p streaming service in mobile/WLAN environment
– Related to PPSP: PPSP can act as the enabler protocols once
3GPP identifies interfaces between components
Conclusion
• PPSP is necessary
• PPSP can be standardized by starting with
– Tracker protocol: Peer and tracker communication
– Peer protocol: Peer communication
• PPSP is supported and to be supported by
some participants who are willing to take part
in the PPSP standards
Thanks for your attention!
Q&A?
Appendix 1: UUSee-Advanced P2P+CDN network
Platform Architecture
User can receive data from
both CDN node and other
peers
Super Node layer:
Tranditional CDN Node or
P2P Super Node
System Manage layer:
CMS/Storage/P2P Tracker
Appendix 2: CoolRuc
• Basic Operation
– Mesh-based network (CoolStreaming
based)
• Adapted for MDC and SVC transmission
– Media Distribution
• Data-driven
• Pull-based
– Peers are grouped into “channels”
• Discovery of peers:
– 1st) Contact Rendezvous peer
(centralized node)
• Get Channel List (specifies tracker peers
of each channel)
– 2nd) Contact Tracker peer for a
specific channel
• Is the same peer as the source node
• Provides an initial list of peers joining the
same channel (random list)
– 3rd) Discovery of new peers by
gossiping protocol (“stady state”
operation)
Appendix 3:NAPA-WINE
European research project:
Network Aware P2P Application over WIse NEtworks
(NAPA-WINE, EU FP7 STREP, http://www.napa-wine.eu)
Study and predict the impact of massive usage of
P2P-streaming on the Internet
• Algorithms and prototype
Chunk Distribution
Tree
for a “network aware” P2P
video streaming application,
Awareness
with optimization based on
and
Overlay
Optimization
Management
measurements between
peers and external
Measurements
guidance (IETF ALTO)
Why P2P Streaming?
YouTube daily net loss is $0.48~$1.28 million
Why?
C/S streaming has
bottleneck in scalability
Server capacity
Bandwidth
consumption
Source: RampRate Website