背景介绍

记录共128W条！
 
SELECT cpe_id, COUNT(*) restarts
FROM business_log
WHERE operate_time>='2012-12-05 00:00:00' AND operate_time<'2018-01-05 00:00:00' AND operate_type=3 AND result=0
GROUP BY cpe_id
 
尝试对原SQL语句进行优化后发现，统计速度依旧没有获得满意的提升。单独运行条件查询语句（不包含GROUP BY和COUNT函数）后发现，查询的结果数据量只有6655条，耗时0.825s；加上统计语句后，时间飙升至3s。

原理

mysql 解释：
MySQL说明文档中关于优化GROUP BY的部分指出：The most general way to satisfy a GROUP BY clause is to scan the whole table and create a new temporary table where all rows from each group are consecutive, and then use this temporary table to discover groups and apply aggregate functions (if any)。即，GROUP BY语句会扫描全表并新建一个临时表用来分组存放数据，然后根据临时表中的分组对数据执行聚合函数。现在的问题聚焦在：如果GROUP BY和WHERE在同一个语句中，这个“全表”指的是物理表还是WHERE过滤后的数据集合？
 
SELECT cpe_id, COUNT(*) restarts
FROM (
SELECT cpe_id
FROM business_log
WHERE operate_time>='2012-12-05 00:00:00' AND operate_time<'2018-01-05 00:00:00' AND operate_type=3 AND result=0
) t
GROUP BY cpe_id
---------------------
如上述语句所示，在查询语句外包了一个统计语句。执行结果：0.851s。时间消耗大幅减少！

结论

利用GROUP BY统计大数据时，应当将查询与统计分离，优化查询语句。
-
 
 

实际业务中

原查询过程
SELECT SUM(a.reportcount)reportcount,a.OrganizationId,a.organizationcode,a.organizationname,org.UpperComCode,org.organizationlevel FROM (SELECT COUNT(1) ReportCount, o.OrganizationCode ,o.organizationname, o.id OrganizationId,o.organizationlevel FROM report r JOIN organization o ON r.OrganizationCode = o.OrganizationCode WHERE o.OrganizationCode in ('44010000','44040000','44006600','44008800','44020000','44050000','44060000','44070000','44080000','44090000','44120000','44150000','44160000','44170000','44180000','44190000','44510000','44520000','44530000','44710000','44940000','44950000') and r.ReportTime>'2019/4/25 0:00:00' and r.ReportReasonSubmitCode in ('A10006','A10007') and r.ReportTime<'2019/5/20 0:00:00' and r.LossTime>'2019/4/11 0:00:00' and r.LossTime<'2019/5/6 0:00:00' and r.InsuranceType in (5,6,7,8) AND r.CatasCollectionId=361 GROUP BY o.OrganizationCode,o.organizationname, o.id,o.organizationlevel union all SELECT COUNT(1) ReportCount, o.OrganizationCode,o.organizationname, o.id OrganizationId,o.organizationlevel FROM report r JOIN organization o ON r.OrganizationCode = o.OrganizationCode WHERE o.UpperComCode IN('44010000','44040000','44006600','44008800','44020000','44050000','44060000','44070000','44080000','44090000','44120000','44150000','44160000','44170000','44180000','44190000','44510000','44520000','44530000','44710000','44940000','44950000')and r.ReportTime>'2019/4/25 0:00:00' and r.ReportReasonSubmitCode in ('A10006','A10007') and r.ReportTime<'2019/5/20 0:00:00' and r.LossTime>'2019/4/11 0:00:00' and r.LossTime<'2019/5/6 0:00:00' and r.InsuranceType in (5,6,7,8) AND r.CatasCollectionId=361 GROUP BY o.OrganizationCode,o.organizationname, o.id ,o.organizationlevel union all SELECT COUNT(1) ReportCount , o.UpperComCode OrganizationCode,o.ParentOrgName OrganizationName, o.ParentOrgId OrganizationId,o.organizationlevel FROM report r JOIN organization o ON r.OrganizationCode = o.OrganizationCode WHERE o.UpperComCode IN(SELECT OrganizationCode FROM organization WHERE UpperComCode IN ('44010000','44040000','44006600','44008800','44020000','44050000','44060000','44070000','44080000','44090000','44120000','44150000','44160000','44170000','44180000','44190000','44510000','44520000','44530000','44710000','44940000','44950000' )and r.ReportTime>'2019/4/25 0:00:00' and r.ReportReasonSubmitCode in ('A10006','A10007') and r.ReportTime<'2019/5/20 0:00:00' and r.LossTime>'2019/4/11 0:00:00' and r.LossTime<'2019/5/6 0:00:00' and r.InsuranceType in (5,6,7,8) AND r.CatasCollectionId=361) GROUP BY o.UpperComCode ,o.ParentOrgName, o.ParentOrgId,o.organizationlevel union all SELECT SUM(reportcount) reportcount , organization.UpperComCode OrganizationCode, organization.ParentOrgName OrganizationName,organization.ParentOrgId OrganizationId,organization.OrganizationLevel FROM ( SELECT COUNT(1) reportcount , o.UpperComCode newcode, o.organizationlevel FROM report r JOIN organization o ON r.OrganizationCode = o.OrganizationCode WHERE o.UpperComCode IN( SELECT OrganizationCode FROM organization WHERE UpperComCode IN(SELECT OrganizationCode FROM organization WHERE UpperComCode IN ('44010000','44040000','44006600','44008800','44020000','44050000','44060000','44070000','44080000','44090000','44120000','44150000','44160000','44170000','44180000','44190000','44510000','44520000','44530000','44710000','44940000','44950000' )and r.ReportTime>'2019/4/25 0:00:00' and r.ReportReasonSubmitCode in ('A10006','A10007') and r.ReportTime<'2019/5/20 0:00:00' and r.LossTime>'2019/4/11 0:00:00' and r.LossTime<'2019/5/6 0:00:00' and r.InsuranceType in (5,6,7,8) AND r.CatasCollectionId=361 ) ) GROUP BY o.UpperComCode, o.organizationlevel ) a JOIN organization ON a.newcode = organization.OrganizationCode GROUP BY organization.UpperComCode , organization.ParentOrgName ,organization.ParentOrgId ,organization.OrganizationLevel ) a JOIN organization org ON a.organizationcode=org.organizationcode GROUP BY a.organizationcode,a.organizationname,org.UpperComCode,org.organizationlevel,a.OrganizationId ORDER BY OrganizationLevel
优化结果
SELECT COUNT(1) AS reportcount,a.OrganizationId,a.organizationcode,a.organizationname,a.UpperComCode,a.organizationlevel FROM (
 
 
SELECT r.*,r.id AS reportid, o.OrganizationCode ,o.organizationname, o.id OrganizationId,o.organizationlevel ,o.UpperComCode
FROM report r JOIN organization o ON r.OrganizationCode = o.OrganizationCode
WHERE o.OrganizationCode IN ('44010000','44040000','44006600','44008800','44020000','44050000','44060000','44070000','44080000','44090000','44120000','44150000','44160000','44170000','44180000','44190000','44510000','44520000','44530000','44710000','44940000','44950000') AND r.ReportTime>'2010/4/25 0:00:00' AND r.ReportReasonSubmitCode IN ('A10006','A10007') AND r.ReportTime<'2019/5/20 0:00:00' AND r.LossTime>'2010/4/11 0:00:00' AND r.LossTime<'2019/5/6 0:00:00' AND r.InsuranceType IN (5,6,7,8) AND r.CatasCollectionId=0
 
UNION ALL
SELECT r.id AS reportid , o.OrganizationCode,o.organizationname, o.id OrganizationId,o.organizationlevel ,o.UpperComCode
FROM report r JOIN organization o ON r.OrganizationCode = o.OrganizationCode
WHERE o.UpperComCode IN('44010000','44040000','44006600','44008800','44020000','44050000','44060000','44070000','44080000','44090000','44120000','44150000','44160000','44170000','44180000','44190000','44510000','44520000','44530000','44710000','44940000','44950000')AND r.ReportTime>'2010/4/25 0:00:00' AND r.ReportReasonSubmitCode IN ('A10006','A10007') AND r.ReportTime<'2019/5/20 0:00:00' AND r.LossTime>'2010/4/11 0:00:00' AND r.LossTime<'2019/5/6 0:00:00' AND r.InsuranceType IN (5,6,7,8) AND r.CatasCollectionId=0
 
UNION ALL
SELECT r.id AS reportid , o.UpperComCode OrganizationCode,o.ParentOrgName OrganizationName, o.ParentOrgId OrganizationId,ou.organizationlevel ,ou.UpperComCode
FROM report r JOIN organization o ON r.OrganizationCode = o.OrganizationCode
JOIN organization ou ON o.UpperComCode=ou.OrganizationCode
WHERE o.UpperComCode IN
(SELECT OrganizationCode FROM organization WHERE UpperComCode IN
('44010000','44040000','44006600','44008800','44020000','44050000','44060000','44070000','44080000','44090000','44120000','44150000','44160000','44170000','44180000','44190000','44510000','44520000','44530000','44710000','44940000','44950000' )AND r.ReportTime>'2010/4/25 0:00:00' AND r.ReportReasonSubmitCode IN ('A10006','A10007') AND r.ReportTime<'2019/5/20 0:00:00' AND r.LossTime>'2010/4/11 0:00:00' AND r.LossTime<'2019/5/6 0:00:00' AND r.InsuranceType IN (5,6,7,8) AND r.CatasCollectionId=0)
 
) a
GROUP BY a.OrganizationCode,a.organizationname, a.OrganizationId,a.organizationlevel,a.UpperComCode
ORDER BY a.organizationlevel
 

mysql 之 group by 性能优化 查询与统计分离的更多相关文章

  1. Web 性能优化： 使用 Webpack 分离数据的正确方法

    摘要: Webpack骚操作. 原文:Web 性能优化: 使用 Webpack 分离数据的正确方法 作者:前端小智 Fundebug经授权转载,版权归原作者所有. 制定向用户提供文件的最佳方式可能是一 ...

  2. MySQL查询语句执行过程及性能优化-查询过程及优化方法（JOIN/ORDER BY）

    在上一篇文章MySQL查询语句执行过程及性能优化-基本概念和EXPLAIN语句简介中介绍了EXPLAIN语句,并举了一个慢查询例子:

  3. MySQL有关Group By的优化

    昨天我写了有关MySQL的loose index scan的相关博文(http://www.cnblogs.com/wingsless/p/5037625.html),后来我发现上次提到的那个优化方法 ...

  4. Mysql Join语法以及性能优化

    引言 内外联结的区别是内联结将去除所有不符合条件的记录,而外联结则保留其中部分.外左联结与外右联结的区别在于如果用A左联结B则A中所有记录都会保留在结果中,此时B中只有符合联结条件的记录,而右联结相反 ...

  5. mysql问题排查与性能优化

     MySQL 问题排查都有哪些手段? 使用 show processlist 命令查看当前所有连接信息. 使用 explain 命令查询 SQL 语句执行计划. 开启慢查询日志,查看慢查询的 SQL. ...

  6. mysql之数据库添加索引优化查询效率

    项目中如果表中的数据过多的话,会影响查询的效率,那么我们需要想办法优化查询,通常添加索引就是我们的选择之一: 1.添加PRIMARY KEY(主键索引) mysql>ALTER TABLE `t ...

  7. mysql use index （） 优化查询的例子

    USE INDEX在你查询语句中表名的后面,添加 USE INDEX 来提供你希望 MySQ 去参考的索引列表,就可以让 MySQL 不再考虑其他可用的索引.Eg:SELECT * FROM myta ...

  8. mysql explain的使用（优化查询）

    explain显示了mysql如何使用索引来处理select语句以及连接表.可以帮助选择更好的索引和写出更优化的查询语句. 1.创建数据库 创建的sql语句如下: /* Navicat MySQL D ...

  9. SqlServer性能优化 查询和索引优化（十二）

    查询优化的过程: 查询优化: 功能:分析语句后最终生成执行计划 分析:获取操作语句参数 索引选择 Join算法选择 创建测试的表: select * into EmployeeOp from Adve ...

随机推荐

  1. Phalcon学习-model

    Model:表与表之间的关系:hasOne 一对一( $fields, $referenceModel, $referencedFields : 当前表中的字段, 对应关系模型, 对应关系模型中表的字 ...

  2. MySQL和Redis面试题小结

    MySQL专题 1. 主键 超键 候选键 外键 主 键: 数据库表中对储存数据对象予以唯一和完整标识的数据列或属性的组合.一个数据列只能有一个主键,且主键的取值不能缺失,即不能为空值(Null). 超 ...

  3. sysctl -p 重新加载文件/etc/sysctl.conf -a 所有参数 -w 临时指定

    sysctl命令用于运行时配置内核参数,这些参数位于/proc/sys目录下.sysctl配置与显示在/proc/sys目录中的内核参数．可以用sysctl来设置或重新设置联网功能,如IP转发.IP碎 ...

  4. SSM 整合

    --- 分为三层: DAO层:负责与数据源进行交互 Service:业务处理层,也可称为服务层,对上层提供统一接口的服务. Controller: 控制器层,负责处理来自客户端的请求. 通用配置: d ...

  5. Spark 基础及RDD基本操作

    什么是RDD RDD(Resilient Distributed Dataset)叫做分布式数据集,是Spark中最基本的数据抽象,它代表一个不可变.可分区.里面的元素可并行计算的集合.RDD具有数据 ...

  6. IntelliJ IDEA 14 利用JRebel实现热部署

    特别鸣谢:http://wlb.wlb.blog.163.com/blog/static/467413201522095132658/ ©IntelliJ IDEA开源社①群 185441009 鸣谢 ...

  7. HttpComponents 基础接口/类与HTTP message的对应关系

    前提一: 什么是HTTP Message -- 遵循HTTP协议发送的消息!其格式是固定的:HTTP Message = Message Line + Message Header + Message ...

  8. hdu 5044 树区间操作最后输出/ lca+dfs

    题意:一棵树,俩种操作:1 有路径上的全部点加vi,2全部边加vi. 先离线求出全部询问的lca,再遍历询问一次,点+vi,lca-2*vi ,最后dfs从叶子扫上来一次,最后再祖先点补上就可以.用了 ...

  9. c#后台访问接口

    直接上代码 后台代码 //接口地址string url = "http://spherefg.topsmoon.com:6666/restapi/Comment/SubmitCommentF ...

  10. svn merge和branch分析

    [转载] 使用svn几年了,一直对分支和合并敬而远之,一来是因为分支的管理不该我操心,二来即使涉及到分支的管理,也不敢贸然使用合并功能,生怕合并出了问题对团队造成不良影响,最主要的原因是,自己对分支的 ...