The Impala Query Language and Built-in Functions - Learning Cloudera Impala - page 81

Loading data from HDFS using the

LOAD DATA statement

As you know, data is stored in HDFS and Impala processes this data. So, when you
need to perform some Extract Transform Load ( ETL ) activity to load the data from
HDFS to Impala tables, you can use LOAD DATA statements. The key properties of
LOAD DATA statements are as follows:

• The loaded data files are moved from HDFS to the Impala data directory
• You can either give a file name from HDFS or a directory name to load all the
files into an Impala table; however, a wild card pattern is not supported with
the HDFS path

The LOAD DATA statement and examples are as follows:

LOAD DATA INPATH 'hdfs_file_or_directory_path'
[OVERWRITE]

INTO TABLE tablename
[PARTITION (partcol1=val1,
partcol2=val2 ...)]

Examples:

CREATE TABLE students (id int, name string);
LOAD DATA INPATH '/user/avkash/students.txt'
INTO TABLE students;

In the previous example, you have to make sure that the students.txt file is loc-
ated at HDFS in folder /user/avkash .