Ranger plugin overview
Ranger enforces authorization using a plugin model.
Ranger at the core has a centralized web application, which consists of the policy administration. These policies are enforced within the Hadoop ecosystem using lightweight Ranger Java plugins. These plugins run as part of the same process as the namenode (HDFS), HiveServer2(Hive), HiveMetaStore, HBase server (Hbase), Kafka, Solr, NiFi, Raz, RazS3, ADLS, Yarn and Knox server (Knox). Plugins are enabled by default for each of these components except (Solr) and can be disabled individually, using Cloudera Manager.
Ranger plugins exist in the path of the user request. Each plugin decides whether to allow or deny user requests for accessing. Each plugin also collects and stores the access request details as access audit log records.
Ranger plugins enforce the policies defined in the policy database. Ranger Admin users can create a policy for a specific set of resources and assign a specific set of permissions to a specific set of users, groups and roles. Ranger admin users manage policies using the Ranger Admin Web UI.
Ranger policies are independent from native permissions (os permission). Ranger uses native permissions to authorize user access in the case that an applicable Ranger policy does not exist in the policy database.