Defining a File-Based Writable External Table
Writable external tables that output data to files use the HAWQ parallel file server program,
gpfdist, or HAWQ Extensions Framework (PXF).
CREATE WRITABLE EXTERNAL TABLE command to define the external table and specify the location and format of the output files.
- With a writable external table using the
gpfdistprotocol, the HAWQ segments send their data to
gpfdist, which writes the data to the named file.
gpfdistmust run on a host that the HAWQ segments can access over the network.
gpfdistpoints to a file location on the output host and writes data received from the HAWQ segments to the file. To divide the output data among multiple files, list multiple
gpfdistURIs in your writable external table definition.
- A writable external web table sends data to an application as a stream of data. For example, unload data from HAWQ and send it to an application that connects to another database or ETL tool to load the data elsewhere. Writable external web tables use the
EXECUTEclause to specify a shell command, script, or application to run on the segment hosts and accept an input stream of data. See Defining a Command-Based Writable External Web Table for more information about using
EXECUTEcommands in a writable external table definition.
You can optionally declare a distribution policy for your writable external tables. By default, writable external tables use a random distribution policy. If the source table you are exporting data from has a hash distribution policy, defining the same distribution key column(s) for the writable external table improves unload performance by eliminating the requirement to move rows over the interconnect. If you unload data from a particular table, you can use the
LIKE clause to copy the column definitions and distribution policy from the source table.