HDFS is distributed and also has redundancy. In particular, you discuss how portions of the data are replicated among DataNodes. This allows for both redundancy and distributed processing especially with map reduce jobs. The DataNodes are assigned to work on separate portions of the replicated data. You would need to diagram the master services like the NameNode, Resource Manager, and Application Master. You could show the master services help distribute the data and map reduce tasks to the Data Nodes.