基于Hadoop与FFmpeg的多媒体分布式处理系统的设计

Design of Multimedia Distributed Processing System based on Hadoop and FFmpeg

视频媒体已经成为一种重要的网络资源,如何快速地为用户提供视频点播支持成为了一个严峻的问题。基于Hadoop分布式文件处理系统和FFmpeg音视频处理技术,构建一套分布式视频转码系统,将任意编码的视频文件转化H.264编码格式,能够实现在线实时点播;利用云端实现了用户转码请求监控、转码任务队列实时调度、运行资源分配等服务。实验证明:该系统的分布式处理有效的提高转码效率,也为资源提供了安全的分布式存储环境。

With video media becoming important network resources, how to quickly provide users with video - on -demand（VOD） support has become a serious problem. Based on Hadoop distributed file processing system and FFmpeg audio and video processing technology, the paper constructs a distributed video transcoding system which is capable of converting the arbitrary encoded video files to H. 264 formats, reaalizing the online real- time VOD service. Meanwhile, utilizing the cloud, the system can achieve the monitoring of user transcoding requests, real - time scheduling of transcoding task queue and allocating of running resources. Experiments have shown that the distributed processing of the system can effectively improve transcoding efficiency and offer safe distributed storage environment to resources as well.