Large data calculation process -hadoop-MapReduce

• MapReduce computation processes

　　1 is the first line of local programmers by MR written by remote command to commit or submit IDE

　　2 a MR program is a Job, Job information will Resourcemanger, registration information to Resourcemanger

• 3 After registering through, Job will copy the relevant resource information (from the HDFS)

  4 Then will submit a complete Job information (including resource information) to Resourcemanger

  5a Resourcemanger by Job information will be submitted to calculate the resources needed to Job, allocate resources for the Job Container

  5b computing resources will be distributed to the corresponding NodeManger, NodeManager creates a MRAppMaster

  6 MRAppMaster initialization Job

  7 obtains input information sliced

  8 MRAppMaster request resources from ResourceManager

  9a starts computing resources (connection to the resource is located corresponding to the NodeManager)

  9b start YARN Child

  10 For complete information from the file system Job

  11 start corresponding Maptask or ReduceTask process, perform calculations.