User Behavior-Based Load Balancing Algorithm for Distributed Streaming Systems

LIN Guangguo,Dai Qionghai,DING Rong
DOI: https://doi.org/10.3321/j.issn:1000-0054.2005.04.026
2005-01-01
Abstract:The behavior characteristics of video-on-demand users is used to balance the system content load distribution. The segment access intensity, which indicates the I/O manipulation frequency of the segment, is defined in terms of the inter-movie and intra-movie skewness. A content-skewness-based load balance algorithm was developed to distribute I/O jobs among the heterogeneous nodes. The algorithm uses parallel streaming servers to divide movies into small segments and to then store them across multiple server nodes according to their capabilities. Simulation test results verify the algorithm's effectiveness with the different definitions of segment access intensity in terms of the mean response time.
What problem does this paper attempt to address?