TY - JOUR
T1 - Cloud-assisted crowdsourced livecast
AU - Zhang, Cong
AU - Liu, Jiangchuan
AU - Wang, Haiyang
PY - 2017/6
Y1 - 2017/6
N2 - The past two years have witnessed an explosion of a new generation of livecast services, represented by Twitch.tv, GamingLive, and Dailymotion, to name but a few. With such a livecast service, geo-distributed Internet users can broadcast any event in real-time, for example, game, cooking, drawing, and so on, to viewers of interest. Its crowdsourced nature enables rich interactions among broadcasters and viewers but also introduces great challenges to accommodate their great scales and dynamics. To fulfill the demands from a large number of heterogeneous broadcasters and geo-distributed viewers, expensive server clusters have been deployed to ingest and transcode live streams. Yet our Twitch-based measurement shows that a significant portion of the unpopular and dynamic broadcasters are consuming considerable system resources; in particular, 25% of bandwidth resources and 30% of computational capacity are used by the broadcasters who do not have any viewers at all. In this article, through the real-world measurement and data analysis, we show that the public cloud has great potentials to address these scalability challenges. We accordingly present the design of Cloud-assisted Crowdsourced Livecast (CACL) and propose a comprehensive set of solutions for broadcaster partitioning. Our trace-driven evaluations show that our CACL design can smartly assign ingesting and transcoding tasks to the elastic cloud virtual machines, providing flexible and costeffective system deployment.
AB - The past two years have witnessed an explosion of a new generation of livecast services, represented by Twitch.tv, GamingLive, and Dailymotion, to name but a few. With such a livecast service, geo-distributed Internet users can broadcast any event in real-time, for example, game, cooking, drawing, and so on, to viewers of interest. Its crowdsourced nature enables rich interactions among broadcasters and viewers but also introduces great challenges to accommodate their great scales and dynamics. To fulfill the demands from a large number of heterogeneous broadcasters and geo-distributed viewers, expensive server clusters have been deployed to ingest and transcode live streams. Yet our Twitch-based measurement shows that a significant portion of the unpopular and dynamic broadcasters are consuming considerable system resources; in particular, 25% of bandwidth resources and 30% of computational capacity are used by the broadcasters who do not have any viewers at all. In this article, through the real-world measurement and data analysis, we show that the public cloud has great potentials to address these scalability challenges. We accordingly present the design of Cloud-assisted Crowdsourced Livecast (CACL) and propose a comprehensive set of solutions for broadcaster partitioning. Our trace-driven evaluations show that our CACL design can smartly assign ingesting and transcoding tasks to the elastic cloud virtual machines, providing flexible and costeffective system deployment.
KW - Crowdsourced livecast
KW - Public clouds
KW - Resource allocation
KW - Workload migration
UR - http://www.scopus.com/inward/record.url?scp=85026482916&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85026482916&partnerID=8YFLogxK
U2 - 10.1145/3095755
DO - 10.1145/3095755
M3 - Article
AN - SCOPUS:85026482916
SN - 1551-6857
VL - 13
JO - ACM Transactions on Multimedia Computing, Communications and Applications
JF - ACM Transactions on Multimedia Computing, Communications and Applications
IS - 3s
M1 - 46
ER -