• 【Oracle】求历史表以往记录中最接近给定时间的一批记录


    需求:某系统将诸多服务器状态定时存到在一张表里,查询时会给出一个时间,要求取出最接近于给定时间,又不可大于的一批记录。

    比如,有gp、dws、etl三台服务器,8:00,9:00,10:00,11:00都有记录存储了它们的状态,当给定10:30时,就该取出10:00的那批记录。

    存储数据的表结构如下:

    create table gpcc_disk_history(
        id number(8),
        hostname nvarchar2(10),
        disksize number(8),
        used number(8),
        ctime date,
        primary key(id)
    )

    测试数据如下:

    insert into gpcc_disk_history values(1,'gp','100','50',to_timestamp('22-10-2021 12:00:00','dd-mm-yyyy hh24:mi:ss'));
    insert into gpcc_disk_history values(2,'dws','100','50',to_timestamp('22-10-2021 12:00:00','dd-mm-yyyy hh24:mi:ss'));
    insert into gpcc_disk_history values(3,'etl','100','50',to_timestamp('22-10-2021 12:00:00','dd-mm-yyyy hh24:mi:ss'));
    
    insert into gpcc_disk_history values(4,'gp','110','51',to_timestamp('22-10-2021 11:00:00','dd-mm-yyyy hh24:mi:ss'));
    insert into gpcc_disk_history values(5,'dws','110','51',to_timestamp('22-10-2021 11:00:00','dd-mm-yyyy hh24:mi:ss'));
    insert into gpcc_disk_history values(6,'etl','110','51',to_timestamp('22-10-2021 11:00:00','dd-mm-yyyy hh24:mi:ss'));
    
    insert into gpcc_disk_history values(7,'gp','120','52',to_timestamp('22-10-2021 10:00:00','dd-mm-yyyy hh24:mi:ss'));
    insert into gpcc_disk_history values(8,'dws','120','52',to_timestamp('22-10-2021 10:00:00','dd-mm-yyyy hh24:mi:ss'));
    insert into gpcc_disk_history values(9,'etl','120','52',to_timestamp('22-10-2021 10:00:00','dd-mm-yyyy hh24:mi:ss'));
    
    insert into gpcc_disk_history values(10,'gp','130','53',to_timestamp('22-10-2021 09:00:00','dd-mm-yyyy hh24:mi:ss'));
    insert into gpcc_disk_history values(11,'dws','130','53',to_timestamp('22-10-2021 09:00:00','dd-mm-yyyy hh24:mi:ss'));
    insert into gpcc_disk_history values(12,'etl','130','53',to_timestamp('22-10-2021 09:00:00','dd-mm-yyyy hh24:mi:ss'));
    
    insert into gpcc_disk_history values(13,'gp','140','54',to_timestamp('22-10-2021 08:00:00','dd-mm-yyyy hh24:mi:ss'));
    insert into gpcc_disk_history values(14,'dws','140','54',to_timestamp('22-10-2021 08:00:00','dd-mm-yyyy hh24:mi:ss'));
    insert into gpcc_disk_history values(15,'etl','140','54',to_timestamp('22-10-2021 08:00:00','dd-mm-yyyy hh24:mi:ss'));

    现在如果要写SQL去查询不大于今天10:30的最新记录,方案是不止一种的,比如:

    --max(ctime)方案  Cost=6
    select id,hostname,disksize,used,to_char(ctime,'hh24:mi:ss') as nctime from gpcc_disk_history where ctime=(select max(ctime) from gpcc_disk_history where ctime<to_timestamp('22-10-2021 10:30:00','dd-mm-yyyy hh24:mi:ss'))

    -- with as max(ctime)方案 Cost=7
    with tmp as (
    select * from gpcc_disk_history where ctime<to_timestamp('22-10-2021 10:30:00','dd-mm-yyyy hh24:mi:ss')
    )
    select id,hostname,disksize,used,to_char(ctime,'hh24:mi:ss') as nctime
    from tmp where ctime=(select max(ctime) as maxtime from tmp)

    --left join方案 Cost=9
    with tmp as (
    select * from gpcc_disk_history where ctime<to_timestamp('22-10-2021 10:30:00','dd-mm-yyyy hh24:mi:ss')
    )
    select a.id,a.hostname,a.disksize,a.used,to_char(a.ctime,'hh24:mi:ss') as nctime
    from tmp a
    left join tmp b
    on a.ctime<b.ctime
    where b.ctime is NULL

    --not exists方案 Cost=9
    with tmp as (
    select * from gpcc_disk_history where ctime<to_timestamp('22-10-2021 10:30:00','dd-mm-yyyy hh24:mi:ss')
    )
    select a.id,a.hostname,a.disksize,a.used,to_char(a.ctime,'hh24:mi:ss') as nctime
    from tmp a
    where not exists (select NULL from tmp b where b.ctime>a.ctime)

    它们执行的结果都是:

            ID HOSTNAME               DISKSIZE       USED NCTIME
    ---------- -------------------- ---------- ---------- ----------------
             8 dws                         120         52 10:00:00
             7 gp                          120         52 10:00:00
             9 etl                         120         52 10:00:00

     从效率和可读性来说,都是第一种方案最优。

    END

  • 相关阅读:
    医学影像
    阿里云九卿 大数据产业化
    陈海青 阿里
    店铺高频问题主动生成知识点机器 大脑+人脑 知识库
    数据总线和流计算在城市大脑中的应用
    无推荐不APP
    ww
    业务架构
    jd算法大赛 一个user_id只需映射到一个sku_id, 但是一个sku_id能否映射到多个user_id
    短URL DH 密钥交换算法
  • 原文地址:https://www.cnblogs.com/heyang78/p/15440654.html
Copyright © 2020-2023  润新知