Hive and Impala data

Hive and Impala data generally reside on HDFS. Hive and Impala replication enables you to copy your Hive metastore and data from one cluster to another. You can synchronize the Hive metastore and data on the destination cluster with the source, based on a specified replication policy.

There are two mechanisms to replicate Hive data on HDFS. The mechanism you choose depends on whether your tables are defined as managed or external.

The following table lists the table type and the replication method you can use to replicate data:


Table Type	Replication Style
Hive 3 Managed Transactional	Hive built-in Replication
Hive 3 External	Replication Manager Hive replication policies

note

Replication Manager currently copies the Hive schema and data details. It does not copy the surrounding metadata and authorization policy in Apache Ranger and Apache Atlas. If you require to synchronize this information, you must do this manually. Both Apache Ranger and Apache Atlas provide import and export APIs to facilitate this. Refer to their respective sections below for more information.

In cases where Hive and Impala utilize non-HDFS storage backends to manage data, such as Kudu or Phoenix on HBase, you must utilize the underlying storage mechanism to manage replication directly. If the underlying storage mechanism has no-native replication capability, such as Kudu, then dual-ingest solutions can be used to keep disaster recovery environments in sync.

For information about using Replication Manager to create Hive replication policies, see Replicating Hive and Impala.