How to create Spark federated tables in MySQL

You can use the SQL Gateway to configure a MySQL remoting service and set up federated tables for Spark data. The service is a daemon process that provides a MySQL interface to the CData ODBC Driver for Spark: After you have started the service, you can create a server and tables using the FEDERATED Storage Engine in MySQL. You can then work with Spark data just as you would local MySQL tables.

Connect to Spark Data

If you have not already done so, provide values for the required connection properties in the data source name (DSN). You can use the built-in Microsoft ODBC Data Source Administrator to configure the DSN. This is also the last step of the driver installation. See the "Getting Started" chapter in the help documentation for a guide to using the Microsoft ODBC Data Source Administrator to create and configure a DSN.

Set the Server, Database, User, and Password connection properties to connect to SparkSQL.

Configure the SQL Gateway

See the SQL Gateway Overview to set up connectivity to Spark data as a virtual MySQL database. You will configure a MySQL remoting service that listens for MySQL requests from clients. The service can be configured in the SQL Gateway UI.

Create a FEDERATED Server and Tables for Spark Data

After you have configured and started the service, create a FEDERATED server to simplify the process of creating FEDERATED tables:

Create a FEDERATED Server

The following statement will create a FEDERATED server based on the ODBC Driver for Spark. Note that the username and password of the FEDERATED server must match a user account you defined on the Users tab of the SQL Gateway.

CREATE SERVER fedSparkSQL
FOREIGN DATA WRAPPER mysql
OPTIONS (USER 'sql_gateway_user', PASSWORD 'sql_gateway_passwd', HOST 'sql_gateway_host', PORT ####, DATABASE 'CData SparkSQL Sys');

Create a FEDERATED Table

To create a FEDERATED table using our newly created server, use the CONNECTION keyword and pass the name of the FEDERATED server and the remote table (Customers). Refer to the following template for the statement to create a FEDERATED table:

CREATE TABLE fed_customers (
  ...,
  city  TYPE(LEN),
  balance  TYPE(LEN),
  ...,
)
ENGINE=FEDERATED
DEFAULT CHARSET=latin1
CONNECTION='fedSparkSQL/customers';

NOTE: The table schema for the FEDERATED table must match the remote table schema exactly. You can always connect directly to the MySQL remoting service using any MySQL client and run a SHOW CREATE TABLE query to get the table schema.

Execute Queries

You can now execute queries to the Spark FEDERATED tables from any tool that can connect to MySQL, which is particularly useful if you need to JOIN data from a local table with data from Spark. Refer to the following example:

SELECT 
  fed_customers.city, 
  local_table.custom_field 
FROM 
  local_table 
JOIN 
  fed_customers 
ON 
  local_table.foreign_city = fed_customers.city;