We had to specify a URL that starts with 'jdbc:google:rdbms', and .Driver as the driver class. Then we override a couple of methods to achieve behavior that is specific to Google Cloud SQL. We extended from the original MySQLDatabaseMeta class to get the default behavior of a MySQL connection type. To create a database plugin for Kettle, we need to implement the DatabaseInterface. To solve the above problems, we decided to create our own database plugin.įor more information on developing your own Kettle database pluginĬreating database plugins at Pentaho InfoCenter. This prevents Kettle from proceeding to display the table in its UI. Unfortunately, calling PreparedStatement.getMetaData on Google Cloud SQL JDBC driver always returns null (even after the statement is executed).

Kettle optimizes by using meta data from a JDBC prepared statement (not from the result set).

But it still fails in retrieving meta information about tables/columns/datatypes.

Using a 'Generic database' connection type allows Kettle to connect successfully.

We needed to use a URL that starts with 'jdbc:google:rdbms://'. It was prefixing the URL with 'jdbc:mysql://'.

Using a MySQL connection type does not allow a custom URL.

But we could not get it to use the correct URL. We tried using the MySQL connection type. We were surprised to see that Kettle could not connect properly to Google Cloud SQL (even if Google says it's running MySQL 5.5). We opted to use Pentaho Data Integration (aka Kettle) to do some ETL. We were playing around Google Cloud SQL and wanted to see if we can migrate some of our MySQL database data onto the cloud.

Pentaho data integration url