Hudi 0.9.0 adds Spark SQL DDL and DML statements for using Hudi, making it easier for all users (including non-engineers or analysts) to access and operate Hudi.

Constraints

You can use Spark SQL to operate Hudi on the Hudi client.
You can use Spark SQL to operate Hudi in JDBCServer of Spark2x.
You cannot use Spark SQL to operate Hudi on the Spark2x client.
You cannot operate Hudi in the Hive or Hetu engine.
The default value of KeyGenerator in SQL is org.apache.hudi.keygen.ComplexKeyGenerator. Therefore, you need to set the KeyGenerator value to that of SQL when data is written in DataSource mode.
If you run the Show Partitions command for the first time, you need to run the msck repair command as prompted.
Spark SQL does not support incremental view query.
Spark SQL does not support ALTER CHANGE DATA TYPE.
To write data incrementally using SQL into a COW table created using DataSource, you need to run the ALTER TABLE table_name SET SERDEPROPERTIES(primaryKey='xx',preCombineField='xx') statement in SQL to reset two properties of the table. To write data incrementally using SQL into an MOR table created using DataSource, you need to specify a path to the one set for creating this MOR table.