I would like to know if it is possible to mix Isilon and nodes with DAS (Direct Attached Storage) in HDFS instantiation in the cluster, for example, with the multiple namespace mechanism (HDFS federation). If this is not possible, is HDP then able to have 2 separate HDFS instantiations in 1 Hadoop cluster (e.g., 1 for Isilon and 1 for the nodes with DAS) while still having one pool of nodes for processing (which may be the nodes with DAS)? I know that applications can specify the uri of the NameNode when submitting jobs to Hadoop. Can this be used for having 2 HDFS instantiations? If so, how does this work in practice? Is Ambari able to manage such a configuration? Do applications need to explicitly refer to NameNodes by uri, or can they refer to tiers instead? Are Hive and Spark able to simultaneously data on both the HDFS instantiations (for example for joining a table on Isilon with a table stored on DAS)? Any help would be appreciated ... View more