项目中统计SQL执行缓慢的方案-数据预处理

使用场景：

由于表数据量巨大，导致一些统计相关的sql执行非常慢，使用户有非常不好的体验，并且sql和数据库已经没有优化空间了。(并且该统计信息数据实时性要求不高的前提下)

解决方案：

整体思路：创建预处理表——通过定时任务将数据插入到结果表——统计信息时直接通过结果表进行查询——大大提高响应速度

注：1.结果表中需要包含查询条件里的所有字段

2.定时任务可以根据实际需要设定频率

3.最好创建一个与结果表表结构一样的临时表用于数据暂存,防止在插入数据这个时间段上导致结果表数据空白。（直接将临时表数据插入到结果表速度很快这段时间可以忽略）

方案示例：

第一步：创建结果表

drop table if exists `user_study_statistics`;

CREATE TABLE `project_statistics` (

  `project_id` varchar(128) CHARACTER SET utf8mb4 COLLATE utf8mb4_bin DEFAULT NULL COMMENT '项目id',

  `project_unit_id` int(11) NOT NULL COMMENT '单位工程id',

  `unit_name` varchar(255) DEFAULT NULL COMMENT '单位工程名称',

  `project_name` varchar(255) DEFAULT NULL COMMENT '项目名称',

  `file_count` int(8) NOT NULL DEFAULT '0' COMMENT '文件总数',

  `file_count_30` int(8) NOT NULL DEFAULT '0' COMMENT '文件总数（30天内）',

  `file_count_90` int(8) NOT NULL DEFAULT '0' COMMENT '文件总数（90天内）',

  `file_count_180` int(8) NOT NULL DEFAULT '0' COMMENT '文件总数（180天内）',

  `check_count` int(8) DEFAULT NULL COMMENT '检查次数',

  `check_count_30` int(8) DEFAULT NULL COMMENT '检查次数（30天内）',

  `check_count_90` int(8) DEFAULT NULL COMMENT '检查次数（90天内）',

  `check_count_180` int(8) DEFAULT NULL COMMENT '检查次数（180天内）',

  `check_num` int(8) NOT NULL DEFAULT '0' COMMENT '检查文件总数',

  `check_num_30` int(8) NOT NULL DEFAULT '0' COMMENT '检查文件总数（30天内）',

  `check_num_90` int(8) NOT NULL DEFAULT '0' COMMENT '检查文件总数（90天内）',

  `check_num_180` int(8) NOT NULL DEFAULT '0' COMMENT '检查文件总数（180天内）',

  `check_scale` varchar(26) DEFAULT NULL COMMENT '检查文件占比',

  `check_scale_30` varchar(26) DEFAULT NULL COMMENT '检查文件占比（30天内）',

  `check_scale_90` varchar(26) DEFAULT NULL COMMENT '检查文件占比（90天内）',

  `check_scale_180` varchar(26) DEFAULT NULL COMMENT '检查文件占比（180天内）',

  `pass_scale` varchar(26) DEFAULT NULL COMMENT '通过率',

  `pass_scale_30` varchar(26) DEFAULT NULL COMMENT '通过率（30天内）',

  `pass_scale_90` varchar(26) DEFAULT NULL COMMENT '通过率（90天内）',

  `pass_scale_180` varchar(26) DEFAULT NULL COMMENT '通过率（180天内）',

  `correct_count` int(8) DEFAULT NULL COMMENT '整改条数',

  `correct_count_30` int(8) DEFAULT NULL COMMENT '整改条数（30天内）',

  `correct_count_90` int(8) DEFAULT NULL COMMENT '整改条数（90天内）',

  `correct_count_180` int(8) DEFAULT NULL COMMENT '整改条数（180天内）',

  `out_time_count` int(8) DEFAULT NULL COMMENT '逾期未完成整改条数',

  `out_time_count_30` int(8) DEFAULT NULL COMMENT '逾期未完成整改条数（30天内）',

  `out_time_count_90` int(8) DEFAULT NULL COMMENT '逾期未完成整改条数（90天内）',

  `out_time_count_180` int(8) DEFAULT NULL COMMENT '逾期未完成整改条数（180天内）'

) ENGINE=InnoDB DEFAULT CHARSCREATE TABLE `project_statistics_tmp` (

  `project_id` varchar(128) CHARACTER SET utf8mb4 COLLATE utf8mb4_bin DEFAULT NULL COMMENT '项目id',

  `project_unit_id` int(11) NOT NULL COMMENT '单位工程id',

  `unit_name` varchar(255) DEFAULT NULL COMMENT '单位工程名称',

  `project_name` varchar(255) DEFAULT NULL COMMENT '项目名称',

  `file_count` int(8) NOT NULL DEFAULT '0' COMMENT '文件总数',

  `file_count_30` int(8) NOT NULL DEFAULT '0' COMMENT '文件总数（30天内）',

  `file_count_90` int(8) NOT NULL DEFAULT '0' COMMENT '文件总数（90天内）',

  `file_count_180` int(8) NOT NULL DEFAULT '0' COMMENT '文件总数（180天内）',

  `check_count` int(8) DEFAULT NULL COMMENT '检查次数',

  `check_count_30` int(8) DEFAULT NULL COMMENT '检查次数（30天内）',

  `check_count_90` int(8) DEFAULT NULL COMMENT '检查次数（90天内）',

  `check_count_180` int(8) DEFAULT NULL COMMENT '检查次数（180天内）',

  `check_num` int(8) NOT NULL DEFAULT '0' COMMENT '检查文件总数',

  `check_num_30` int(8) NOT NULL DEFAULT '0' COMMENT '检查文件总数（30天内）',

  `check_num_90` int(8) NOT NULL DEFAULT '0' COMMENT '检查文件总数（90天内）',

  `check_num_180` int(8) NOT NULL DEFAULT '0' COMMENT '检查文件总数（180天内）',

  `check_scale` varchar(26) DEFAULT NULL COMMENT '检查文件占比',

  `check_scale_30` varchar(26) DEFAULT NULL COMMENT '检查文件占比（30天内）',

  `check_scale_90` varchar(26) DEFAULT NULL COMMENT '检查文件占比（90天内）',

  `check_scale_180` varchar(26) DEFAULT NULL COMMENT '检查文件占比（180天内）',

  `pass_scale` varchar(26) DEFAULT NULL COMMENT '通过率',

  `pass_scale_30` varchar(26) DEFAULT NULL COMMENT '通过率（30天内）',

  `pass_scale_90` varchar(26) DEFAULT NULL COMMENT '通过率（90天内）',

  `pass_scale_180` varchar(26) DEFAULT NULL COMMENT '通过率（180天内）',

  `correct_count` int(8) DEFAULT NULL COMMENT '整改条数',

  `correct_count_30` int(8) DEFAULT NULL COMMENT '整改条数（30天内）',

  `correct_count_90` int(8) DEFAULT NULL COMMENT '整改条数（90天内）',

  `correct_count_180` int(8) DEFAULT NULL COMMENT '整改条数（180天内）',

  `out_time_count` int(8) DEFAULT NULL COMMENT '逾期未完成整改条数',

  `out_time_count_30` int(8) DEFAULT NULL COMMENT '逾期未完成整改条数（30天内）',

  `out_time_count_90` int(8) DEFAULT NULL COMMENT '逾期未完成整改条数（90天内）',

  `out_time_count_180` int(8) DEFAULT NULL COMMENT '逾期未完成整改条数（180天内）'

) ENGINE=InnoDB DEFAULT CHARSET=utf8 COMMENT='项目检查状态统计表';

ET=utf8 COMMENT='项目检查状态统计表_临时表';

创建视图：

CREATE VIEW `v_project_statistics_info` AS

SELECT project_id,project_unit_id,unit_name,project_name,

COUNT(1) file_count,-- 文件总数

COUNT((datediff(date(now()),date(gmt_create)) <=30 ) OR null) file_count_30,

COUNT((datediff(date(now()),date(gmt_create)) <=90 ) OR null) file_count_90,

COUNT((datediff(date(now()),date(gmt_create)) <=180 ) OR null) file_count_180,

SUM(check_count) as check_count,-- 检查次数

SUM(CASE WHEN (datediff(date(now()),date(modify_time)) <=30 ) THEN check_count ELSE 0 END) AS check_count_30,

SUM(CASE WHEN (datediff(date(now()),date(modify_time)) <=90 ) THEN check_count ELSE 0 END) AS check_count_90,

SUM(CASE WHEN (datediff(date(now()),date(modify_time)) <=180 ) THEN check_count ELSE 0 END) AS check_count_180,

COUNT(check_count>0 OR null) as check_num,-- 检查文件总数

COUNT((datediff(date(now()),date(modify_time)) <=30 ) and check_count>0 OR null) as check_num_30,

COUNT((datediff(date(now()),date(modify_time)) <=90 ) and check_count>0 OR null) as check_num_90,

COUNT((datediff(date(now()),date(modify_time)) <=180 ) and check_count>0 OR null) as check_num_180,

CONCAT(ROUND(COUNT(check_count>0 OR null)/COUNT(1)*100,0),'%') as check_tage, -- 检查文件占比

CONCAT(ROUND(COUNT((datediff(date(now()),date(modify_time)) <=30 ) and check_count>0 OR null)/COUNT((datediff(date(now()),date(gmt_create)) <=30 ) OR null)*100,0),'%') as check_scale_30,

CONCAT(ROUND(COUNT((datediff(date(now()),date(modify_time)) <=90 ) and check_count>0 OR null)/COUNT((datediff(date(now()),date(gmt_create)) <=90 ) OR null)*100,0),'%') as check_scale_90,

CONCAT(ROUND(COUNT((datediff(date(now()),date(modify_time)) <=180 ) and check_count>0 OR null)/COUNT((datediff(date(now()),date(gmt_create)) <=180 ) OR null)*100,0),'%') as check_scale_180,

CONCAT(ROUND(COUNT(check_status=2 OR null)/COUNT(check_count>0 OR null)*100,0),'%') as pass_tage, -- 通过率

CONCAT(ROUND(COUNT((datediff(date(now()),date(modify_time)) <=30 ) and check_status=2 OR null)/COUNT((datediff(date(now()),date(modify_time)) <=30 ) and check_count>0 OR null)*100,0),'%') as pass_scale_30,

CONCAT(ROUND(COUNT((datediff(date(now()),date(modify_time)) <=90 ) and check_status=2 OR null)/COUNT((datediff(date(now()),date(modify_time)) <=90 ) and check_count>0 OR null)*100,0),'%') as pass_scale_90,

CONCAT(ROUND(COUNT((datediff(date(now()),date(modify_time)) <=180 ) and check_status=2 OR null)/COUNT((datediff(date(now()),date(modify_time)) <=180 ) and check_count>0 OR null)*100,0),'%') as pass_scale_180,

SUM(correct_count) as correct_count,-- 整改条数

SUM(correct_count_30) as correct_count_30,

SUM(correct_count_90) as correct_count_90,

SUM(correct_count_180) as correct_count_180,

SUM(out_time_count) as out_time_count, -- 逾期未完成整改条数

SUM(out_time_count_30) as out_time_count_30,

SUM(out_time_count_90) as out_time_count_90,

SUM(out_time_count_180) as out_time_count_180

from (

SELECT SUBSTRING_INDEX(d.bus_id,'_',1) as project_id,d.project_unit_id,d.node_id,ifnull(d.check_count,0) as check_count,d.check_status,ifnull(a.correct_count,0) as correct_count,ifnull(a.correct_count_30,0) as correct_count_30,ifnull(a.correct_count_90,0) as correct_count_90,ifnull(a.correct_count_180,0) as correct_count_180,ifnull(a.out_time_count,0) as out_time_count,ifnull(a.out_time_count_30,0) as out_time_count_30,ifnull(a.out_time_count_90,0) as out_time_count_90,ifnull(a.out_time_count_180,0) as out_time_count_180,u.unit_name,p.project_name,d.modify_time,d.gmt_create from v_doc_info d

INNER JOIN project_detail p on p.local_project_id=d.project_id

LEFT JOIN project_unit u on u.id=d.project_unit_id

left join(

SELECT local_project_id,IFNULL(correct_soure_id,0)as node_id,COUNT(1) as correct_count

,COUNT(datediff(date(now()),date(create_time)) <=30) as correct_count_30

,COUNT(datediff(date(now()),date(create_time)) <=90) as correct_count_90

,COUNT(datediff(date(now()),date(create_time)) <=180) as correct_count_180

,count((have_deleted=0 and correct_end_time<NOW()) or null) as out_time_count

,count((have_deleted=0 and correct_end_time<NOW() and (datediff(date(now()),date(create_time)) <=30)) or null) as out_time_count_30

,count((have_deleted=0 and correct_end_time<NOW() and (datediff(date(now()),date(create_time)) <=90)) or null) as out_time_count_90

,count((have_deleted=0 and correct_end_time<NOW() and (datediff(date(now()),date(create_time)) <=180)) or null) as out_time_count_180  from doc_correct GROUP BY local_project_id,correct_soure_id

) a on a.node_id=d.node_id

UNION ALL

SELECT d.project_id,d.remote_project_id as project_unit_id,d.id as node_id,ifnull(d.check_count,0) as check_count,d.check_status,ifnull(a.correct_count,0) as correct_count,ifnull(a.correct_count_30,0) as correct_count_30,ifnull(a.correct_count_90,0) as correct_count_90,ifnull(a.correct_count_180,0) as correct_count_180,ifnull(a.out_time_count,0) as out_time_count,ifnull(a.out_time_count_30,0) as out_time_count_30,ifnull(a.out_time_count_90,0) as out_time_count_90,ifnull(a.out_time_count_180,0) as out_time_count_180,u.unit_name,p.project_name,d.modify_time,d.gmt_create from v_datumdata_info d

INNER JOIN project_detail p on p.local_project_id=d.project_id

LEFT JOIN project_unit u on u.id=d.remote_project_id

left join(

SELECT local_project_id,IFNULL(correct_soure_id,0)as node_id,COUNT(1) as correct_count

,COUNT(datediff(date(now()),date(create_time)) <=30) as correct_count_30

,COUNT(datediff(date(now()),date(create_time)) <=90) as correct_count_90

,COUNT(datediff(date(now()),date(create_time)) <=180) as correct_count_180

,count((have_deleted=0 and correct_end_time<NOW()) or null) as out_time_count

,count((have_deleted=0 and correct_end_time<NOW() and (datediff(date(now()),date(create_time)) <=30)) or null) as out_time_count_30

,count((have_deleted=0 and correct_end_time<NOW() and (datediff(date(now()),date(create_time)) <=90)) or null) as out_time_count_90

,count((have_deleted=0 and correct_end_time<NOW() and (datediff(date(now()),date(create_time)) <=180)) or null) as out_time_count_180  from doc_correct GROUP BY local_project_id,correct_soure_id

) a on a.node_id=d.id

)t GROUP BY project_id,project_unit_id

第二步：创建预处理定时任务

    /**

     * 资料检查统计预加载定时任务

     */

    @PostConstruct //程序启动时就执行一次

    @Scheduled(cron = "${schedul.statistics.checkStatistics:0 0 * * * ?}") //默认定时频率每小时执行一次

    public void preProjectStatistics(){

        //0.删除临时表里的数据

        statisticsDao.deleteStatisticsTmpData();

        //1.将数据插入临时表

        statisticsDao.insertStatisticsTmpData();

        //2.删除正式表里的数据

        statisticsDao.deleteStatisticsData();

        //3.将临时表里的数据插入正式表

        statisticsDao.insertStatisticsData();

        //3.删除临时表里的数据

        statisticsDao.deleteStatisticsTmpData();

    }

对应的sql语句：

<insert id="insertStatisticsTmpData">

        insert into project_statistics_tmp SELECT * from v_project_statistics_info

    </insert>

    <delete id="deleteStatisticsData" parameterType="java.lang.Long">

        delete from project_statistics;

    </delete>

    <insert id="insertStatisticsData">

        insert into project_statistics

    select * from project_statistics_tmp;

    </insert>

    <delete id="deleteStatisticsTmpData" parameterType="java.lang.Long">

        delete from project_statistics_tmp;

    </delete>

第三步：修改之前的统计sql从多表关联到查询单表速度有了质的提升

博客园排版可能有点乱，可转至我个人链接地址：https://raokun.top/raokun/archives/项目中统计sql执行缓慢的方案-数据预处理

项目中统计SQL执行缓慢的方案-数据预处理

项目中统计SQL执行缓慢的方案-数据预处理的相关教程结束。

相关推荐

C语言之预处理命令的示例分析

SElinux 导致 Keepalived 检测脚本无法执行

你真的知道吗？catch、finally和return哪个先执行

执行pod setup 报错error: RPC failed; curl 18 transfer closed with outstanding read data remainin

django中执行py报错Requested setting DEFAULT_INDEX_TABLESPACE, but settings are not configured

[转]Mysql中的SQL优化与执行计划

pytest的几种执行方式

Apache log4j2 远程代码执行漏洞复现👻