Embed Size (px)
DESCRIPTION
To effectively serve massive volumes of video traffic content delivery networks (CDNs) are turning to clients for assistance, creating hybrid peer-assisted content delivery systems. In this project we analyze how peer-assisted CDNs are affected by a number of design obstacles which include: the need of localizing peer-to-peer traffic within ISPs (isp-friendliness), reluctance of users to participate in redistributing the content (partial participation) and necessity to match users with similar bitrate requirements (bitrate stratification)
Citation preview
CD-GAIN: Measuring Traffic Gains from Peer-assisted Content Delivery
Dmytro Karamshuk1, Nishanth Sastry1, Jigna Chandaria2, Andy Secker2
1King's College London, 2BBC R&D
ESPRC project in collaboration with BBC R&D
Video Delivery Network user
44% of UK' households
user user user
user user
user
Video traffic now accounts for 66% of all traffic on the Internet and is predicted to reach 79% by 2018
over 50% of all traffic for streaming TV
Appetite comes with Eating
BBC iPlayer
Content is King
Video Delivery Network
user
user user
user
useruser
average of 5K users online every sec in the first day after release
5K duplicates
Solution: ask users for assistance
Higher Demand means Higher Savings
ci=ui r iAverage number of peers online
G(c)=1−(c(1+ec c−m
(m Γ(m)−Γ(1+m , c)))m
+1)
−1
How much we can offload to peers
Model traffic swarms as infinite-server queue(extending Menasche et al., CoNext 2009)
The share of traffic which can be offloaded to peers is growing with the demand and the size of content
request ratecontent duration
London, Sep. 2013
Users - 2.2 M IP address - 1.3 M Sessions - 15.9 M Traffic - 6.5 Gbps
Records in the Dataset:<network id, ISP, user id, request time, session duration,
bitrate, content id>
So, How Much we can Offload?
theo
retic
al
● model is in a good agreement with simulations● significant traffic savings can be achieved despite constraints
What do we do with the ObstaclesEight biggest ISPs account for
over 70% of trafficTwo main bitrate formats
dominate in over 70% of sessions
Fragmentation of traffic by ISP and bitrate formats yields several large content swarms
Top-5% of the content corpus accounts for 80% of traffic
Why it works in the end?
● “online while watching” model is important
● most traffic is generated by a small amount of popular content
Takeaways● online while watching model and long-duration content are crucial factors for feasibility of peer-assistant content delivery
● fragmentation of peer-to-peer swarms by ISP, quality format and content doesn't impose significant limitations on the system
● up to 80% of all traffic for the whole content corpus in iPlayer can be offloaded for the biggest ISPs
Visit our website: http://bit.ly/cd-gain
Dmytro KaramshukKing's College London
follow me on Twitter: @karamshuk
![[Streamroot] Whitepaper peer assisted adaptive streaming](https://img.dokumen.tips/doc/110x75/587fe8b81a28ab46228b5da3/streamroot-whitepaper-peer-assisted-adaptive-streaming.jpg)
