Apache Spark ODBC

Type of Support:	Read & Write; In-Database
Validated On:	Apache Spark 1.2.0; Simba Apache Spark Driver 1.02.04.1005
Connection Type:	ODBC (32- and 64-bit)
Driver Details:	The ODBC driver can be downloaded here. In-Database processing requires 64-bit database drivers.
Driver Configuration Requirements:	For optimal performance, you must enable the Fast SQLPrepare option within the driver Advanced Options to allow Alteryx to retrieve metadata without running a query.

Alteryx tools used to connect

• Input Data Tool and Output Data Tool (Standard workflow processing)
• Connect In-DB Tool and Data Stream In Tool (In-database workflow processing)

Additional Details

To use the Apache Spark ODBC, you must have Apache Spark SQL enabled. Not all Hadoop distributions support Apache Spark. If you are unable to connect using Apache Spark ODBC, contact your Hadoop vendor for instructions on how to set up the Apache Spark server correctly.

If you have issues with reading or writing Unicode® characters, access the Simba Impala ODBC driver. Under Advanced Options, select the “Use SQL Unicode Types” option.

Read Support

Install and configure the Apache Spark ODBC driver:

• Spark Server Type: Select the appropriate server type for the version of Apache Spark that you are running. If you are running Apache Spark 1.1 and later, then select Apache SparkThriftServer.
• Authentication Mechanism: See the installation guide downloaded with the Simba Apache Spark driver to configure this setting based on your setup.

To set up the driver Advanced Options, see the installation guide downloaded with the Simba Apache Spark driver.

Write Support

• For both standard and in-database workflows, use the Data Stream In Tool to write to Apache Spark. Write support is via HDFS.
• If you are writing with HDFS Avro, you must select the Default WebHDFS (50070) port option in the HDFS Avro Connection properties window.

To write a table with field names that total more than 4000 characters, use CSV instead of Avro.