Load or replace partition data feature

There is no difference in the way you insert data into a partitioned or unpartitioned Iceberg table.

Working with partitions is easy because you write the query in the same way for the following operations:

  • Insert into, or replace, an unpartitioned table
  • Insert into, or replace, an identity partitioned table
  • Insert into, or replace, a transform-partitioned table

Do not use INSERT OVERWRITE on tables that went through partition evolution. Truncate such tables first, and then INSERT the tables.

Impala example

CREATE TABLE ice_12 (i int, s string, t timestamp, t2 timestamp) STORED BY ICEBERG;

Hive or Impala examples

INSERT INTO ice_12 VALUES (42, 'impala', now(), to_date(now())); 
INSERT OVERWRITE ice_t VALUES (42, 'impala', now(), to_date(now()));