本節(jié)哼绑，我們?yōu)榇蠹医榻B下JDBC的批處理問題泪喊。

向student表中插入一萬條數(shù)據(jù)

package com.iotek.advance;

import java.sql.Connection;
import java.sql.PreparedStatement;
import java.sql.SQLException;

import com.iotek.common.ConnectionFactory;
import com.iotek.common.UtilsJdbc;

public class BatchTest {
    public static void main(String[] args) {
        
        Connection connection = null;
        PreparedStatement preparedStatement = null;
        int count = 0;
        
        try {
            connection = ConnectionFactory.getConnection();
            connection.setAutoCommit(false);
            String insertSQL = "INSERT INTO student VALUES (?,?,?,?,?)";
            preparedStatement = connection.prepareStatement(insertSQL);
            
            long time1 = System.currentTimeMillis();
            for (int i = 0; i < 10000; i++) {
                preparedStatement.setInt(1, i+1);
                preparedStatement.setString(2, "zhangsan" + i);
                preparedStatement.setString(3, "beijing" + i);
                preparedStatement.setString(4, "male");
                preparedStatement.setInt(5, 20);
                preparedStatement.executeUpdate();
                count++;
            }
            connection.commit();
            
            long time2 = System.currentTimeMillis();
            System.out.println("insert " + count + " rows takes " + (time2 - time1) + " ms");
            
        } catch (SQLException e) {
            e.printStackTrace();
            try {
                connection.rollback();
            } catch (SQLException e1) {
                e1.printStackTrace();
            }
        } finally {
            UtilsJdbc.close(connection, preparedStatement, null);
        }
    }
}

運行結果為：2556 ms

采用批處理方式插入數(shù)據(jù)，關鍵方法是addBatch()和executeBatch()

package com.iotek.advance;

import java.sql.Connection;
import java.sql.PreparedStatement;
import java.sql.SQLException;

import com.iotek.common.ConnectionFactory;
import com.iotek.common.UtilsJdbc;

public class BatchTest {
    public static void main(String[] args) {
        
        Connection connection = null;
        PreparedStatement preparedStatement = null;
        int count = 0;
        
        try {
            connection = ConnectionFactory.getConnection();
            connection.setAutoCommit(false);
            String insertSQL = "INSERT INTO student VALUES (?,?,?,?,?)";
            preparedStatement = connection.prepareStatement(insertSQL);
            
            long time1 = System.currentTimeMillis();
            for (int i = 0; i < 10000; i++) {
                preparedStatement.setInt(1, i+1);
                preparedStatement.setString(2, "zhangsan" + i);
                preparedStatement.setString(3, "beijing" + i);
                preparedStatement.setString(4, "male");
                preparedStatement.setInt(5, 20);
                preparedStatement.addBatch();
                count++;
            }
            
            preparedStatement.executeBatch();
            connection.commit();
            
            long time2 = System.currentTimeMillis();
            System.out.println("insert " + count + " rows takes " + (time2 - time1) + " ms");
            
        } catch (SQLException e) {
            e.printStackTrace();
            try {
                connection.rollback();
            } catch (SQLException e1) {
                e1.printStackTrace();
            }
        } finally {
            UtilsJdbc.close(connection, preparedStatement, null);
        }
    }
}

運行結果為：166 ms

由此搀突，我們可以看到批處理僅僅占用了之前寫法的零頭毫秒锰悼，大大提升了效率柳骄，所以在大量處理數(shù)據(jù)庫數(shù)據(jù)的時候团赏，最后使用批處理箕般，將參數(shù)或者是SQL語句，放到statement維護的緩存中舔清，然后統(tǒng)一丝里，在一次與數(shù)據(jù)庫的連接中，統(tǒng)一提交体谒，這樣可以大大提升性能杯聚。