跨库查询(同一实例下)

用navicat跨库查询,窗口所处的库不影响,因为你带了库名.表名(因为你处在同一实例下)

调用数据迁移的存储过程huike_transfer_hour完成小时表的分库

存储过程完成小时表分库: 主要逻辑是拿到库里的所有年份 2016 2017 2018 遍历游标

DECLARE date_year VARCHAR(20);

DECLARE rs CURSOR FOR SELECT YEAR(datatime) date_year FROM report_site_hour GROUP BY YEAR(datatime);

遍历游标动态拼接sql 更新插入到指定库

SET @table_replace = CONCAT('REPLACE INTO ipvacloud_ramus_',date_year,'.report_site_hour

select * from ipvacloud_20170424.report_site_hour a WHERE YEAR(datatime)=',date_year,'

AND a.modifytime between "',minModifytime,'" AND "',maxModifytime,'"');

PREPARE sql_oper_replace FROM @table_replace;

EXECUTE sql_oper_replace;

ipvacloud_ramus_2016
ipvacloud_ramus_2017

//这里查询的是分库后的小时表如果是2个时间可以用工具类DatabaseTools获取[2016,2017]
List<Integer> list = DatabaseTools.getFromToYears("2016", "2017");
paraMap.put("year", year);

不跨库只查询单年的小时表 SQL里直接写:

select inum from ipvacloud_ramus_${year}.report_site_hour rsh

package com.winner.common;
import java.util.ArrayList;
import java.util.List;
//获取数据库分区,即获取年份list的工具类
public class DatabaseTools {
//获取两个时间中的年份返回list,在数据库遍历[2016,2017]
public static List<Integer> getFromToYears(String from, String to){
Integer start = Integer.valueOf(from);
Integer end = Integer.valueOf(to);
List<Integer> years = new ArrayList<>();
if (start.equals(end)) {
years.add(start);
}else{
do {
years.add(start);
start++;
} while (!start.equals(end));
years.add(end);
}
return years;
}
}

这个子查询里有beginDate,endDate,siteKey,years遍历的item, 跨库查询 ipvacloud_ramus_${year}.report_site_hour
因为跨年了,所以foreach遍历,相当于关联了2个年份的小时表小时表跨年查询的应用场景 foreach带()带别名rsh

left join (

select innum from ipvacloud_ramus_${item}.report_site_hour where datatime <![CDATA[>]]> #{startDate}

and datatime <![CDATA[<]]> #{endDate}

</foreach>
) rsh

如下是foreach部分作为子查询生成的sql,如果不跨年,则不出现union(即子查询是本年的)
(
SELECT
innum
FROM
ipvacloud_ramus_2016.report_site_hour
WHERE
datatime > ?
AND datatime < ?
UNION
SELECT
innum
FROM
ipvacloud_ramus_2017.report_site_hour
WHERE
datatime > ?
AND datatime < ?
) rsh

跨库查询(同一实例下)

猜你喜欢