多数据源的配置

   大型应用为了实现读写分离，便于数据的管理，提高系统的水平伸缩性，往往会用到多数据源。

1，数据源是对应不同数据库的，为了方便操作可以写一个数据库的配置文件

UPDATE_MYSQL.type= mysql

UPDATE_MYSQL.autoCommitOnClose= false

UPDATE_MYSQL.driverClassName= com.mysql.jdbc.Driver

UPDATE_MYSQL.url= jdbc:mysql://192.168.1.139:3306/testdatabase?useUnicode=true&characterEncoding=UTF-8&useOldAliasMetadataBehavior=true

UPDATE_MYSQL.username= xccmysql

UPDATE_MYSQL.password= xccW7W8

UPDATE_MYSQL.initialPoolSize= 5

SELECT_MYSQL.type= mysql

SELECT_MYSQL.autoCommitOnClose= false

SELECT_MYSQL.driverClassName= com.mysql.jdbc.Driver

SELECT_MYSQL.url= jdbc:mysql://192.168.1.139:3306/testmuti?useUnicode=true&characterEncoding=UTF-8&useOldAliasMetadataBehavior=true

SELECT_MYSQL.username= root

SELECT_MYSQL.password= xcc2009081231

SELECT_MYSQL.initialPoolSize= 5

2，然后再spring的配置文件里添加不同的数据源

<!-- 数据源的配置 -->

      < bean id = "update_dataSource" class = "org.apache.commons.dbcp.BasicDataSource" destroy-method = "close" >

           < property name = "driverClassName" value = "${UPDATE_MYSQL.driverClassName}" />

           < property name = "url" value = "${UPDATE_MYSQL.url}" />

           < property name = "username" value = "${UPDATE_MYSQL.username}" />

           < property name = "password" value = "${UPDATE_MYSQL.password}" />

           <!-- data source configuration -->

           < property name = "initialSize" value = "${UPDATE_MYSQL.initialPoolSize}" /> <!-- initial connections -->

           < property name = "maxActive" value = "10" /> <!-- MAX connections -->

           < property name = "maxIdle" value = "10" /> <!-- MAX idle connections -->

           < property name = "minIdle" value = "5" /> <!-- MIN idle connections -->

           <!-- 处理mysql 8小时自动断开连接的问题 -->

           < property name = "testWhileIdle" value = "true" />

           < property name = "testOnBorrow" value = "false" />

           < property name = "testOnReturn" value = "false" />

           < property name = "validationQuery" value = "select 1" />

           < property name = "timeBetweenEvictionRunsMillis" value = "20000" />

           < property name = "numTestsPerEvictionRun" value = "100" />

      </ bean >

     

     

      < bean id = "select_dataSource" class = "org.apache.commons.dbcp.BasicDataSource" destroy-method = "close" >

           < property name = "driverClassName" value = "${UPDATE_MYSQL.driverClassName}" />

           < property name = "url" value = "${SELECT_MYSQL.url}" />

           < property name = "username" value = "${SELECT_MYSQL.username}" />

           < property name = "password" value = "${SELECT_MYSQL.password}" />

           <!-- data source configuration -->

           < property name = "initialSize" value = "${SELECT_MYSQL.initialPoolSize}" /> <!-- initial connections -->

           < property name = "maxActive" value = "10" /> <!-- MAX connections -->

           < property name = "maxIdle" value = "10" /> <!-- MAX idle connections -->

           < property name = "minIdle" value = "5" /> <!-- MIN idle connections -->

           <!-- 处理mysql 8小时自动断开连接的问题 -->

           < property name = "testWhileIdle" value = "true" />

           < property name = "testOnBorrow" value = "false" />

           < property name = "testOnReturn" value = "false" />

           < property name = "validationQuery" value = "select 1" />

           < property name = "timeBetweenEvictionRunsMillis" value = "20000" />

           < property name = "numTestsPerEvictionRun" value = "100" />

      </ bean >

3，数据源准备好之后，基本工作，已完成，先放在一边，我们去java里面写数据源的名称常量类

package org.iwall.muti.datasource;

/**

 *

 * @author Lee

 *

 */

public class DataSourceNames {

      public static final String UPDATE_MYSQL = "update_dataSource" ;

      public static final String SELECT_MYSQL = "select_dataSource" ;

}

4，建立一个类，负责改变数据源的名称（其实可使用spring的注解切换不同的数据源）

package org.iwall.muti.datasource;

import org.apache.commons.logging.Log;

import org.apache.commons.logging.LogFactory;

/**

 *  获得和设置上下文环境 主要负责改变上下文数据源的名称

 * @author Lee

 *

 */

public class DataSourceContextHolder {

    private static final ThreadLocal contextHandler = new ThreadLocal(); // 线程本地环境

    public static Log log = LogFactory.getLog(DataSourceContextHolder.class);

    // 设置数据源类型

    public static void setDataSourceType(String dataSourceType) {

        log.info("set dataSource:" + dataSourceType);

        contextHandler.set(dataSourceType);

    }

    // 获取数据源类型 　　

    public static String getDataSourceType() {

        return (String) contextHandler.get();

    }

    // 清除数据源类型

    public static void clearDataSourceType() {

        contextHandler.remove();

    }

}

5，建立动态数据源，主要实现 determineCurrentLookupKey这个方法

package org.iwall.muti.datasource;

import java.util.Map;

import org.apache.commons.logging.Log;

import org.apache.commons.logging.LogFactory;

import org.springframework.beans.factory.InitializingBean;

import org.springframework.jdbc.datasource.lookup.AbstractRoutingDataSource;

import org.springframework.jdbc.datasource.lookup.DataSourceLookup;

/**

 *

 * @author Lee

 *

 */

public class DynamicDataSource extends AbstractRoutingDataSource implements InitializingBean {

    private Log log = LogFactory.getLog(DynamicDataSource.class);

    /* 

     * 该方法必须要重写  方法是为了根据数据库标示符取得当前的数据库 既是获得数据源类型

     */ 

    @Override 

    public Object determineCurrentLookupKey() { 

        log.info("starting determineCurrentLookupKey:" + DataSourceContextHolder.getDataSourceType());

        return DataSourceContextHolder.getDataSourceType();

    } 

    @Override 

    public void setDataSourceLookup(DataSourceLookup dataSourceLookup) { 

        log.info("starting setDataSourceLookup:" + DataSourceContextHolder.getDataSourceType());

        super.setDataSourceLookup(dataSourceLookup); 

    } 

    @Override 

    public void setDefaultTargetDataSource(Object defaultTargetDataSource) {

        log.info("starting setDefaultTargetDataSource:" + DataSourceContextHolder.getDataSourceType());

        super.setDefaultTargetDataSource(defaultTargetDataSource); 

    } 

    @Override 

    public void setTargetDataSources(Map targetDataSources) {

        System.out.println("starting setTargetDataSources:" + DataSourceContextHolder.getDataSourceType());

        super.setTargetDataSources(targetDataSources); 

        //重点 

        super.afterPropertiesSet(); 

    }

}

6，数据源配置完了，java对应的类也写完了，就需要写映射关系了，其中key是对应java类里的数据源常量名，value是配置文件里的数据源名

< bean id = "dynamicDataSource" class = "org.iwall.muti.datasource.DynamicDataSource" >

           <!-- 通过key-value的形式来关联数据源 -->

           < property name = "targetDataSources" >

               < map key-type = "java.lang.String" >

                    < entry key = "update_dataSource" value-ref = "update_dataSource" ></ entry >

                    < entry key = "select_dataSource" value-ref = "select_dataSource" ></ entry >

               </ map >

           </ property >

           < property name = "defaultTargetDataSource" ref = "update_dataSource" >

           </ property >

  </ bean >

7，如果没有使用hibernate，mybatis等数据交互的框架，用spring的模板jdbc的话，注入一下就好

< bean id = "jdbcTemplate" class = "org.springframework.jdbc.core.JdbcTemplate" >

           < constructor-arg ref = "dynamicDataSource" />

  </ bean >

8，总结下，这种方式，便于维护，基本只用修改数据库配置文件和spring配置文件即可。但是多个用户访问会争抢datasource，系统性能

会降低，（没测试过）。