Reseach Of The Cloud Mass Video Transcoding Based On Mapreduce Programming Model

At present, the cloud computing technology has become the research hotspotwhich is not only the continuation and development of the parallel computing, thedistributed computing and the grid computing technologies, but also a new-stylecommercial calculation mode. Many large companies, scientific researchinstitutions and colleges and universities have been engaged in the researches inthis field. The cloud computing technology integrates various computing andstorage resources through the internet and thus forms a virtual computing resourcepool and provides resources to users by means of services, making it convenient forpeople to obtain strong computing and storing capabilities and infrastructures. Thecloud computing technology can efficiently solve problems in analyzing andprocessing mass data and supplies with a highly reliable and extensible dataprocessing and storage center, improving the data processing capability andreducing the requirements on the terminal equipments as well. Therefore, thecomputing mode by distributing complex calculations consuming a large quantityof computing resources into multiple nodes through the internet has become anefficient solutionAt present,hadoop is the most widely used, open source distributed computingframework in the area of cloud computing,Hadoop is an open source distributedcomputing framework that is organized by Apache.has been used in theYahoo,Facebook,Baidu,IBM,and other large web application.The most core designof Hadoop computing framework is the programming model of MapReduce andHDFS distributed file system.MapReduce design idea is “the work decompositionand the computer results summary”.HDFS is the file system of hadoop, file storagefor cloud computing.This paper researched the related theory of Cloud Computing,thecharacteristics and advantages of the Cloud Computing.the key technology of Cloud Computing,explored the operating mechanism and principle of the hadoopframework of distributed and open source,then studied the MapReduceprogramming model and its programming method.presents a distributed videotranscoding system framework.At present,hadoop is uses the FIFO algorithm torealize the load balance,task scheduling,it is not conducive to multi-user、multi-tasking,the migration of data exchange will occers when a node failure,itwill take the pressure to the back-end database、disk.This paper compares thevarious,analyzed various scheduling strategies on task execution efficiency.This article about the open source software Hadoop, Mapreduce, HDFS andvideo transcoding software ffmpeg, mencoder combination of research, has formeda set of commercial cloud computing solutions and implementation method. Atpresent our country is the main push triple-play technology, involves the maincontent is the audio and video file processing, this system in the triple playenvironment has great practical significance and application value, extends throughthe upper interface application can be applied to the mobile terminal, the PCterminal, as well as television and radio network.