减少 Postgres SQL 的执行时间

Question

我根据要求写了一个如下的 sql，它给出了正确的结果，但它需要大约 7 秒

select 
  distinct m.mdl, m.src
from model m
join company c on c.name = m.name
  and c.partner = m.partner
where c.id = 1234
and m.status = 'ACTIVE'
EXCEPT 
select
  md.mdl,
  md.src
from model md 
right join (
    select
      distinct m.mdl, m.src
    from model m, company c
    where c.name = m.name
    and c.partner = m.partner 
    and c.id in (123,1122)
) t on md.mdl = t.mdl
  and md.src = t.src
join company cp on cp.name = md.name
  and cp.partner = md.partner
where cp.id = 1234
and md.status = 'ACTIVE'

任何重写的建议，以便将上述 sql 中的执行时间从 7 秒减少。

我尝试使用 WITH 查询编写 SQL，它给出了正确的结果，但需要大约 30 秒。

Answer 1

您的查询看起来像您尝试获取 id 为 1234 的模型，但不是 ids 123&1122。

那些可能会返回相同的结果：

select 
   m.mdl, m.src
from model m
join company c on c.name = m.name
  and c.partner = m.partner
where m.status = 'ACTIVE'
group by m.mdl, m.src
having count(case when c.id = 1234 then 1 end) = 1        -- 1234 exists
   and count(case when c.id in (123,1122) then 1 end) = 0 -- but not 123 or 1122

select 
  distinct m.mdl, m.src
from model m
join company c on c.name = m.name
  and c.partner = m.partner
where c.id = 1234
and m.status = 'ACTIVE'
AND NOT EXISTS
 ( select 1
    from company c
    where c.name = m.name
    and c.partner = m.partner 
    and c.id in (123,1122)
 )