| With the rapid development of the Triple Play,video services are developing rapidly. The need of video for multi-format and multi-platform is becoming more and more urgent. In light of this, the media content operators need to build dedicated video transcoding facilities to meet their demand of transcoding. However, the construction and maintenance of video transcoding facilities needs to invest huge human and financial resources, medium and small scale media content operators lack the conditions in this regard.With the cloud hosting services at low cost, high reliability, easy management features, this paper designs and implements a distributed video transcoding system for medium and small scale media content operators by replacing actual servers with virtual nodes. The system has three types of nodes: DJANGO based scheduling management node, Node.js based video processing node and NFS based file storage node, and the paper designs and implements two transcoding modes: single-step transcoding mode and multi-step transcoding mode on this basis in the system. This paper also optimizes the system on transcoding task selection strategy and node allocation strategy.The former schedules short tasks preferentially in order to reduce their average response time. The latter tries to make time-consuming transcoding tasks are assigned to the high-performance, small load nodes so that it can improve the efficiency of transcoding. Finally, this paper deploys the transcoding system on the Super-Computer cloud platform and shows the effectiveness of transcoding systems by experiments.This paper designs and implements a video transcoding system that can not only meet the transcoding need of medium and small scale media content operators,but also within their affordability on cloud platform which has resource gathering, easy access, and low cost properties. This system has a strong practical significance and application value in the current video industry. |
PDF Full Text Request