Running SQL Statements

Running a Common SQL Statement

To enable an application to operate data in the database by running SQL statements (statements that do not need to transfer parameters), perform the following operations:

1. Create a statement object by calling the createStatement method in Connection.

   1 2	Connection conn = DriverManager.getConnection("url","user","password"); Statement stmt = conn.createStatement();

2. Run the SQL statement by calling the executeUpdate method in Statement.

   1	int rc = stmt.executeUpdate("CREATE TABLE customer_t1(c_customer_sk INTEGER, c_customer_name VARCHAR(32));");

   

   • If an execution request (not in a transaction block) received in the database contains multiple statements, the request is packed into a transaction. The VACUUM operation is not supported in a transaction block. If one of the statements fails, the entire request will be rolled back.

   • Use semicolons (;) to separate statements. Stored procedures, functions, and anonymous blocks do not support multi-statement execution. When preferQueryMode is set to simple, the statement does not execute the parsing logic, and the semicolons (;) cannot be used to separate statements in this scenario.
   • The slash (/) can be used as the terminator for creating a single stored procedure, function, anonymous block, or package body. When preferQueryMode is set to simple, the statement does not execute the parsing logic, and the slash (/) cannot be used as the terminator in this scenario.
   • When prepareThreshold is set to 1, each SQL statement executed by the statement is cached because cached statements are not evicted by default (default value of preferQueryMode). As a result, memory bloat may occur. In this case, set preferQueryMode to extendedCacheEverything to evict cached statements.

3. Close the statement object.

   1	stmt.close();

Running a Prepared SQL Statement

Prepared statements are complied and optimized once but can be used in different scenarios by assigning multiple values. Using prepared statements improves execution efficiency. If you want to run a statement for several times, use a precompiled statement. Perform the following operations:

1. Create a prepared statement object by calling the prepareStatement method in Connection.

   1	PreparedStatement pstmt = con.prepareStatement("UPDATE customer_t1 SET c_customer_name = ? WHERE c_customer_sk = 1");

2. Set parameters by calling the setShort method in PreparedStatement.

   1	pstmt.setShort(1, (short)2);

3. Run the prepared statement by calling the executeUpdate method in PreparedStatement.

   1	int rowcount = pstmt.executeUpdate();

4. Close the prepared statement object by calling the close method in PreparedStatement.

   1	pstmt.close();

Calling a Stored Procedure

To call an existing stored procedure by using JDBC in GaussDB, perform the following operations:

1. Create a call statement object by calling the prepareCall method in Connection.

   1 2	Connection myConn = DriverManager.getConnection("url","user","password"); CallableStatement cstmt = myConn.prepareCall("{? = CALL TESTPROC(?,?,?)}");

2. Set parameters by calling the setInt method in CallableStatement.

   1 2 3

   cstmt.setInt(2, 50); cstmt.setInt(1, 20); cstmt.setInt(3, 90);

3. Register an output parameter by calling the registerOutParameter method in CallableStatement.

   1	cstmt.registerOutParameter(4, Types.INTEGER); // Register an OUT parameter of the integer type.

4. Call the stored procedure by calling the execute method in CallableStatement.

   1	cstmt.execute();

5. Obtain the output parameter by calling the getInt method in CallableStatement.

   1	int out = cstmt.getInt(4); // Obtain the OUT parameter.

   Example:

   1 2 3 4 5 6 7 8 9 10 11 12

   // The following stored procedure (containing the OUT parameter) has been created: create or replace procedure testproc ( psv_in1 in integer, psv_in2 in integer, psv_inout in out integer ) as begin psv_inout := psv_in1 + psv_in2 + psv_inout; end; /

6. Close the call statement by calling the close method in CallableStatement.

   1	cstmt.close();

   

   • Many database classes such as Connection, Statement, and ResultSet have a close() method. Close these classes after using their objects. Closing Connection will close all the related Statements, and closing a Statement will close its ResultSet.
   • Some JDBC drivers support named parameters, which can be used to set parameters by name rather than sequence. If a parameter has the default value, you do not need to specify any parameter value but can use the default value directly. Even though the parameter sequence changes during a stored procedure, the application does not need to be modified. Currently, the GaussDB JDBC driver does not support this method.
   • GaussDB does not support functions containing OUT parameters, or stored procedures and function parameters containing default values.
   • When you bind parameters in myConn.prepareCall("{? = CALL TESTPROC(?,?,?)}") during a stored procedure calling, you can bind parameters according to the placeholders sequence and register the first or the fourth parameter as the output parameter. The preceding example registers the fourth parameter.
   

   • If JDBC is used to call a stored procedure whose returned value is a cursor, the returned cursor cannot be used.
   • A stored procedure and an SQL statement must be run separately.
   • Output parameters must be registered for parameters of the inout type in the stored procedure.

Batch Processing

When a prepared statement processes multiple pieces of similar data, the database creates only one execution plan. This improves compilation and optimization efficiency. Perform the following operations:

1. Create a prepared statement object by calling the prepareStatement method in Connection.

   1 2	Connection conn = DriverManager.getConnection("url","user","password"); PreparedStatement pstmt = conn.prepareStatement("INSERT INTO customer_t1 VALUES (?)");

2. Call the setShort parameter for each piece of data, and call addBatch to confirm that the setting is complete.

   1 2	pstmt.setShort(1, (short)2); pstmt.addBatch();

3. Perform batch processing by calling the executeBatch method in PreparedStatement.

   1	int[] rowcount = pstmt.executeBatch();

4. Close the prepared statement object by calling the close method in PreparedStatement.

   1	pstmt.close();

   

   Do not terminate a batch processing action when it is ongoing; otherwise, database performance will deteriorate. Therefore, disable automatic commit during batch processing. Manually commit several rows at a time. The statement for disabling automatic commit is conn.setAutoCommit(false);.

Adding Single-Shard Execution Syntaxes to Statements

1. Set the nodeName parameter by calling setClientInfo("nodeName","dnx") in Connection.

   Connection conn = getConnection();
   conn.setClientInfo("nodeName","datanode1");

2. Execute the SQL statements by using the executeQuery(String sql) and execute(String sql) methods in Statement and the executeQuery() and execute() methods in PreparedStatement.

   PreparedStatement pstm = conn.prepareStatement("select * from test");
   pstm.execute();
   pstm.executeQuery();

   Statement stmt=conn.createStatement();
   stmt.execute("select * from test");
   stmt.executeQuery("select * from test");

3. Set the parameter to an empty string to disable it.

   conn.setClientInfo("nodeName","");

   

   1. This function is adapted based on the single-shard execution function of the kernel. Therefore, before using this function, check whether the database kernel supports single-shard execution.
   2. After the parameter is enabled, you must manually disable it. Otherwise, the execution of other query statements will be affected.
   3. Once this parameter is enabled, all statements of the current connection will be executed on a specified DN.
   4. After the parameter is enabled, the cache mechanism of PreparedStatement will be affected, cached statements will be cleared, and subsequent statements executed for single-shard queries will not be cached until the parameter is disabled.
   5. The parameter is a connection parameter. Therefore, the parameter value takes effect once. The API cannot be used to execute the statements on different shards at the same time.