如何改进SQL Server中的更新语句

我使用的是SQL Server，我有两个表，第一个叫做MCross

它有3列

Brand1  Brand2  Cnt
--------------------
Mazda   Toyota  NULL
Ford    Toyota  NULL
Nissan  Mazda   NULL
BMW     Opel    NULL
Nissan  BMW     NULL

另一张表叫做Tags

AgentID  Brand     Class
----------------------
1        Toyota    C
1        Toyota    S
1        BMW       X7
1        BMW       X5 
1        BMW       X3
1        Mazda     3
2        Toyota    C
2        Toyota    S
2        Mazda     5
2        Mazda     3
3        BMW       X3
3        Nissan    P
4        Nissan    L
4        BMW       X3

我想用Brand1和Brand2的代理数量更新MCross

我试过这个

但是它非常慢，我认为可以用更好的方式来写

update A
    set Cnt = B.cnt
from MCross A
inner join (
    select C.Brand1, C.Brand2, count(T.AgentID) cnt
    from MCross C
    inner join (
        select AgentID, ','+STRING_AGG(cast(Brand as NVARCHAR(MAX)),',')+',' AS Brands
        from Tags
        group by AgentID
    ) T on T.Brands Like  '%,'+C.Brand1+',%' and T.Brands Like  '%,'+C.Brand2+',%' 
    group by C.Brand1, C.Brand2
) B on A.Brand1 = B.Brand1 and A.Brand2 = B.Brand2

---

看看这是否适合您：

编辑：

嗯。您可以使用应用程序：

注意：如果您想对数据中的所有品牌对进行此操作，只需使用自连接：

with a as (
      select distinct agentid, brand
      from tags
     )
select a1.brand, a2.brand, count(*)
from a a1 join
     a a2
     on a1.agentid = a2.agentid and
        a1.brand <> a2.brand
group by a1.brand, a2.brand;

---

我忘了提到同一品牌可以在同一家代理商重复使用检查新的。

update m
   set m.cnt = a.num_agents
   from mcross m outer apply
        (select count(*) as num_agents
         from (select t.agentid
               from tags t
               where t.brand in (m.brand1, m.brand2)
               group by t.agentid
               having count(distinct t.brand) = 2
              ) a
        ) a;

with a as (
      select distinct agentid, brand
      from tags
     )
select a1.brand, a2.brand, count(*)
from a a1 join
     a a2
     on a1.agentid = a2.agentid and
        a1.brand <> a2.brand
group by a1.brand, a2.brand;