Revision history [back]

The syntax for watermark in SQL used in Databricks is:

CREATE TABLE <table_name>
(<column_1> <data_type_1>,
<column_2> <data_type_2>,
<column_3> <data_type_3>,
…
…
…
<column_n> <data_type_n>)
USING <file_format>
OPTIONS (
  'path'='<path_to_data>',
  'checkpointLocation'='<path_to_checkpoint_directory>',
  'watermark'='<event_time_column_name>','<delay_threshold>'
);

where <table_name> is the name of the table, <column_1>, <column_2>, ..., <column_n> are the names of the columns in the table, <data_type_1>, <data_type_2>, ..., <data_type_n> are the data types of the columns in the table. <file_format> is the file format of the data, <path_to_data> is the path to the data, <path_to_checkpoint_directory> is the path to the checkpoint directory, <event_time_column_name> is the name of the event_time column in the table, and <delay_threshold> is the delay threshold.