Characterizing the service usage of online video sharing system: Uploading vs. playback

Large online video sharing systems are becoming increasingly prevalent among the general public recently. A better understanding of their service usages is crucial for allocating the network resources and adjusting the service design. In this paper, we focus on a leading online video sharing system in China, namely Youku. We collected over 12 billion traffic traces from the network operator, and crawled the video meta-data for 30 consecutive days from Youku website. Based on those data, we present the first comparative characterization work on two key usages of online video sharing system: uploading and playback. We first check the different temporal patterns and intensities of uploaders' and viewers' activities. Then, the properties of the uploaded videos and watched videos are compared, to reveal the various user preferences behind those two service usages. Next, we study the dynamics of video popularity in video uploading and playback, in terms of view count distribution and active period. At last, we look through the followers of uploaders and the network formed by viewers, to analyze the relationships between different kinds of video service users. Our analysis results in this paper can be utilized by network operators, service providers, content creators and other interest parties of online video service, to adjust their service design and improve the user experience.