Study On Multimedia Management Technology Based On Cloud Platform

With the development of Internet technology and the popularization of Internet,the amount of information is exploding,a large number of media are being applied to our economy,society,education, communication, medical, advisory services, and daily life.But in the face of the rapid growth of the amount of information, traditional multimedia management mode can not meet the rapid growth of multimedia resources. So more and more enterprises and units tend to realize the management of multimedia resources on the cloud platform,it provides a more reliable and efficient solution for the management of multimedia information,makes data storage safe and efficient,reduce the overhead of the platform,reduce initial investment,it greatly enhance the utilization of enterprise platform resources.But there are two shortcomings in the existing cloud based multimedia management platform:1) the medium and small file in the multimedia resource library is a very large scale,and the storage of small files and reading is the system’s shortcomings,mass small file will lead to poor file transfer efficiency,and its hard disk read and write efficiency is particularly low.Massive small files will result in the cloud platform to spend more resources for the storage of information,and the process of the document retrieval and reading is also very complex;2) Multimedia resource library’s video often needs to be transcoding to adapt to different resolutions,but the traditional way of transcoding time is long,it requires high node configuration requirements,it will leads node load pressure,can not give users a better user experience.Based on the above requirements, this paper optimizes the management of the traditional multimedia resources from two perspectives.On the one hand, use Word2 Vec and HowNet for multimedia similarity calculation between small files,then according to the results of the clustering algorithm to group all the files,according to the results of the group to achieve the merger of small files, improve the efficiency of the merger of small files;on the other hand, with the task allocation of distributed computing, the distributed transcoding is performed for the multimedia video in the resource pool, and the efficiency of the transcoding is promoted.In order to solve the problem of the multimedia resource management platform,the main results of this paper are as follows:1,Ceph small file definition and identification:Based on the analysis of a large number of experimental data, the paper gets and defines the boundary line of the size file for Ceph, which is used for the identification of small files in the multimedia file upload service.2,Document semantic similarity clustering:In this paper, combined with the characteristics of multimedia files, through semantic similarity computing, clustering, file merging, the small files that are closely related to each other are merged into the same file.3,Distributed transcoding of multimedia video:Taking Ceph as the underlying storage of multimedia video, MapReduce as a platform for computing framework, through the distributed cluster nodes of the ffmpeg tool for distributed video clips, save the time of video transcoding, improve the efficiency of video transcoding, and maximize the use of idle resources.Based on the above research, this paper constructs a prototype of multimedia resource management based on Ceph, and improve related modules, it provides a platform for the future research of distributed file storage and computing.