Because the driver is responsible for routing transactions to the correct grid node, statements must use parameters (rather than literal values) to specify partitioning key values, and the application must provide the required value for the partitioning key. The driver can then use the value that is provided by the application to identify the node that hosts the column value for the partitioning key, and route the transaction to the correct node.

For example, the following statement (where the first column is the partitioning key) should succeed:

INSERT INTO TAB1 VALUES(?, 'TWO HUNDRED');

but the following statement fails (unless the active node connection happens to be to the node that contains the primary partition for the value 200):

INSERT INTO TAB1 VALUES(200, 'TWO HUNDRED');