Hello,    Oracle Data Integrator connects to Hive by using JDBC and uses Hive and the Hive Query Language (HiveQL), a SQL-like language for implementing MapReduce jobs. Source - HERE   The points mentioned by you from the documentation is for the purpose of Blocking the external applications and non service users from accessing the Hive metastore.   Since, ODI connects to Hive using JDBC, it should connect to HiveServer2 as described in this documentation. Once connected, the query executed from ODI will connect with HiveServer2. Then, HiveServer2 will connect with HiveMetastore for getting the metadata details of the table against which you are querying and proceed with the execution. It is not necessary for ODI to connect to Hive MetaStore directly.    For details about Hive Metastore HA, please read HERE ... View more