如何优化我们使用事务板的查询？答案

【问题标题】：How to optimize query where we are using transaction slab?如何优化我们使用事务板的查询？
【发布时间】：2021-12-11 11:45:33
【问题描述】：

我需要根据事务板生成一个老化报告，即在 2、5、10 等间隔上。用户将选择报告开始日期和结束日期，然后我们需要为该特定日期范围生成报告。报告应采用这样的方式，即交易应分成多个板块；例如 0-2、3-4、5-6、6-8 和大于 8 天。我为每个平板编写了一个子查询，它给了我所需的数据。但是由于数据每天都在增加，因此需要花费太多时间。我也尝试了索引，但它并没有提高性能。

select id_ledger as ledger_id,
       customer_id,
       title     as ledger,
       (
           select COALESCE((sum(dr_amount) - sum(clearance_amount)), 0) as diff_value
           from journal_voucher_details
           where cr_amount = 0
             and (('2021-10-23' - transaction_date) <= (2))
             and ledger_id = l.customer_id
             and transaction_date >= '2021-10-01'
             and transaction_date <= '2021-10-24'
           group by ledger_id
       )         as slab1,
       (
           select COALESCE((sum(dr_amount) - sum(clearance_amount)), 0) as diff_value
           from journal_voucher_details
           where cr_amount = 0
             and (('2021-10-23' - transaction_date) <= (2 * 2))
             and (('2021-10-23' - transaction_date) > (2 * 1))
             and ledger_id = l.customer_id
             and transaction_date >= '2021-10-01'
             and transaction_date <= '2021-10-24'
           group by ledger_id
       )         as slab2,
       (
           select COALESCE((sum(dr_amount) - sum(clearance_amount)), 0) as diff_value
           from journal_voucher_details
           where cr_amount = 0
             and (('2021-10-23' - transaction_date) <= (2 * 3))
             and (('2021-10-23' - transaction_date) > (2 * 2))
             and ledger_id = l.customer_id
             and transaction_date >= '2021-10-01'
             and transaction_date <= '2021-10-24'
           group by ledger_id
       )         as slab3,
       (
           select COALESCE((sum(dr_amount) - sum(clearance_amount)), 0) as diff_value
           from journal_voucher_details
           where cr_amount = 0
             and (('2021-10-23' - transaction_date) <= (2 * 4))
             and (('2021-10-23' - transaction_date) > (2 * 3))
             and ledger_id = l.customer_id
             and transaction_date >= '2021-10-01'
             and transaction_date <= '2021-10-24'
           group by ledger_id
       )         as slab4,
       (
           select COALESCE((sum(dr_amount) - sum(clearance_amount)), 0) as diff_value
           from journal_voucher_details
           where cr_amount = 0
             and (('2021-10-23' - transaction_date) > (2 * 4))
             and ledger_id = l.customer_id
             and transaction_date >= '2021-10-01'
             and transaction_date <= '2021-10-24'
           group by ledger_id
       )         as slab5,
       (
           select COALESCE((sum(dr_amount) - sum(clearance_amount)), 0) as diff_value
           from journal_voucher_details
           where cr_amount = 0
             and 1 = 1
             and ledger_id = l.customer_id
             and transaction_date >= '2021-10-01'
             and transaction_date <= '2021-10-24'
           group by ledger_id
       )         as balance
from ledgers l;

注意： 2021-10-1 ==> 开始日期

2021-10-24 ==> 结束日期

2021-10-23 ==> 报告日期

选择2作为板号。

如何提高此查询的性能？

提前致谢。

【问题讨论】：

标签： sql postgresql query-optimization

【解决方案1】：

你可以试试这个，应该更快：

select id_ledger as ledger_id, customer_id, title as ledger,
     , COALESCE( sum(dr_amount) FILTER (WHERE ('2021-10-23' - transaction_date) <= (2)) OVER () 
                 - sum(clearance_amount) FILTER (WHERE ('2021-10-23' - transaction_date) <= (2)) OVER ()
               , 0
               ) as slab1
     , COALESCE(sum(dr_amount) FILTER (where ('2021-10-23' - transaction_date) <= (2 * 2) and ('2021-10-23' - transaction_date) > (2 * 1)) OVER () 
                - sum(clearance_amount) FILTER (where ('2021-10-23' - transaction_date) <= (2 * 2) and ('2021-10-23' - transaction_date) > (2 * 1)) OVER () 
               , 0
               ) as slab2
     , COALESCE(sum(dr_amount) FILTER (where ('2021-10-23' - transaction_date) <= (2 * 3) and ('2021-10-23' - transaction_date) > (2 * 2)) OVER () 
                 - sum(clearance_amount) FILTER (where ('2021-10-23' - transaction_date) <= (2 * 3) and ('2021-10-23' - transaction_date) > (2 * 2)) OVER () 
               , 0
               ) as slab3
     , COALESCE(sum(dr_amount) FILTER (where ('2021-10-23' - transaction_date) <= (2 * 4) and ('2021-10-23' - transaction_date) > (2 * 3)) OVER () 
                 - sum(clearance_amount) FILTER (where ('2021-10-23' - transaction_date) <= (2 * 4) and ('2021-10-23' - transaction_date) > (2 * 3)) OVER () 
               , 0
               ) as slab4
     , COALESCE(sum(dr_amount) FILTER (where ('2021-10-23' - transaction_date) > (2 * 4)) OVER ()
                - sum(clearance_amount) FILTER (where ('2021-10-23' - transaction_date) > (2 * 4)) OVER ()
               , 0
               ) as slab5
     , COALESCE(sum(dr_amount) - sum(clearance_amount), 0) AS balance
from ledgers l
inner join journal_voucher_details j
on j.ledger_id = l.customer_id
where cr_amount = 0
and transaction_date >= '2021-10-01'
and transaction_date <= '2021-10-24'
group by j.ledger_id

【讨论】：

发现查询执行有显着改进。感谢您的帮助。