The Community for Technology Leaders
IEEE International Performance Computing and Communications Conference (2011)
Orlando, FL, USA
Nov. 17, 2011 to Nov. 19, 2011
ISBN: 978-1-4673-0010-0
pp: 1-8
Jun Li , Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
Zhongcheng Li , Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
Haibo Wu , Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
Jing Liu , Inner Mongolia University, Hohhot, China
Yi Sun , Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
Hai Jiang , Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
ABSTRACT
Peer-to-Peer (P2P) technology has been widely adopted by various live streaming systems recently, due to its better scalability and lower costs compared with the client-server architecture. However, P2P live streaming systems are still challenged by the flash crowd scenarios, which often occur when a great number of users suddenly arrive and compete for the limited upload bandwidth of a P2P system. In this case, users are usually subject to a long startup delay and are likely to retry multiple times before leave out of impatience. Current studies mainly focus on the measurement of practical systems and model analysis on flash crowd, but there are few specific approaches so far. In this paper, we develop a capacity-aware user access control algorithm to relieve the flash crowd problem. Firstly, we control the peers to enter the system at a proper rate, which avoids too high arrival rate slowing down the increase of system scale. Secondly, to increase the system service capacity as soon as possible, we let the peers with higher capacity enter the system ahead of the peers with lower capacity. Finally, we also consider the waiting time of peers with low capacity and let them in before they lose patience. To evaluate our algorithm, a new analysis model is also proposed. Simulation experiments and model analysis reveal that our algorithm is more effective to increase the system scale, and can achieve shorter user waiting time as well as lower reject rate.
INDEX TERMS
CITATION
Jun Li, Zhongcheng Li, Haibo Wu, Jing Liu, Yi Sun, Hai Jiang, "How P2P live streaming systems scale quickly under a flash crowd?", IEEE International Performance Computing and Communications Conference, vol. 00, no. , pp. 1-8, 2011, doi:10.1109/PCCC.2011.6108083
176 ms
(Ver 3.3 (11022016))