Time filter

Source Type

Shanghai, China

Yi Z.,Beijing Institute of Technology | Dan H.,Network Security Technology | Yansong Y.,Beijing Institute of Technology | Yan H.,Tencent Research | Changjia C.,Beijing Jiaotong University
IET Conference Publications | Year: 2014

Due to the inefficient resource adjustment, the current P2P file sharing systems cannot achieve the balanced relationship between supply and demand over the resources. Especially after a popular content releasing, a burst of downloaders often can't find sufficient uploaders and their request may starve the upload capacity of server. Therefore the overall system QoS may be degraded. To tackle such issue, this paper proposes a download rate accelerate mechanism, called motivate mechanism. With it, the system can quickly find out the files becoming insufficient by monitoring the operating status of the files hourly. Then it promptly increases the number of copies of those files by using free rider nodes so that the whole system QoS is maintained and the system performance is improved. The experiment results on the practical operating system of Tencent demonstrated that the proposed mechanism increases the download rate, saves the traffic on the server and optimizes the system performance. Source

Yang H.,Beihang University | Luan Z.,Beihang University | Li W.,Beihang University | Qian D.,Beihang University | Guan G.,Tencent Research
Proceedings of the 2012 IEEE 26th International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2012 | Year: 2012

Large-scale data-intensive computing with MapReduce framework in Cloud is becoming pervasive for the core business of many academic, government, and industrial organizations. Hadoop is by far the most successful realization of MapReduce framework. While MapReduce is easy-to-use, efficient and reliable for data-intensive computations, the excessive configuration parameters in Hadoop cause unexpected challenges when running various workloads with Hadoop cluster effectively. Consequently, developers who have less experience with the Hadoop configuration system may devote a significant effort to write an application with poor performance, because they have no idea how these configurations would influence the performance, or they are not even aware that these configurations exist. In this paper, we propose a statistic analysis approach to identify the relationships among workload characteristics, Hadoop configurations and workload performance. Several non-intuitive relationships between workload characteristics and relative performance are revealed and the experimental results demonstrate that our regression models accurately predict the performance of MapReduce workloads under different Hadoop configurations. © 2012 IEEE. Source

Zhang T.,CAS Institute of Computing Technology | Li Z.,University of Science and Technology of China | Shen H.,CAS Institute of Computing Technology | Huang Y.,Tencent Research | Cheng X.,CAS Institute of Computing Technology
Proceedings - International Conference on Computer Communications and Networks, ICCCN | Year: 2011

P2P-VoD systems have gained tremendous popularity in recent years. While existing research is mostly based on theoretical or conventional assumptions, it is particularly valuable to understand and examine how these assumptions work in realistic environments, so as to set up a solid foundation for mechanism design and optimization possibilities. In this paper, we present a comprehensive measurement study of CoolFish, a real-world P2P-VoD system. Our measurement provides several new findings which are different from the traditional assumptions or observations: the access pattern does not match Poisson distribution; session time does not have positive correlation with movie popularity; jump frequency does not have a negative correlation with movie popularity as assumed in previous studies. We analyze the reasons for these results and provide suggestions for the further study of P2P-VoD services. © 2011 IEEE. Source

Zhou Y.,Chinese University of Hong Kong | Fu T.Z.J.,Chinese University of Hong Kong | Chiu D.M.,Chinese University of Hong Kong | Huang Y.,Tencent Research
IEEE Transactions on Multimedia | Year: 2013

Video content downloading using the P2P approach is scalable, but does not always give good performance. Recently, subscription-based premium services have emerged, referred to as cloud downloading. In this service, the cloud storage and server caches user-interested content and updates the cache based on user downloading requests. If a requested video is not in the cache, the request is held in a waiting state until the cache is updated. We call this design server mode. An alternative design is to let the cloud server serve all downloading requests as soon as they arrive, behaving as a helper peer. We call this design helper mode. Our model and analysis show that both these designs are useful for certain operating regimes. The helper mode is good at handling a high request rate, while the server mode is good at scaling with video population size. We design an adaptive algorithm (AMS) to select the service mode automatically. Intuitively, AMS switches service mode from server mode to helper mode when too many peers request blocked movies, and vice versa. The ability of AMS to achieve good performance in different operating regimes is validated by simulation. © 1999-2012 IEEE. Source

He R.,Beihang University | Luan Z.,Beihang University | Huang Y.,Beihang University | Cheng Z.,Beihang University | And 2 more authors.
Proceedings of the 2012 15th International Conference on Network-Based Information Systems, NBIS 2012 | Year: 2012

With the popularity of cloud computing, high volume data needs to be processed in real-time. Therefore, many distributed stream processing systems have risen to deal with this requirement. One kind of distributed stream processing applications contains several PEs which are connected together to implement the application logic. As data is in the form of streams and it needs to be processed in real-time, high availability is important for the application. At the same time, the application requires a high performance, including a low resource cost. According to these features of the application, we propose a new replica placement method to guarantee the application's availability while controlling the resource cost under some constraint and trying to get the maximal resource usage. The result of our experiment shows that our placement method can acquire high availability and low resource cost comparing with other placement strategies. © 2012 IEEE. Source

Discover hidden collaborations