如何计算的总运行在sql向后

如果你的 ID 是独一无二的每一行，然后删除 PARTITION BY ID 否则它不会累积总和. 扭转了正在运行的总数只是反的 ORDER BY 的 SUM:

言:

declare @sales table (
    ID int,
    Date date,
    Amount int);

insert into @sales
values
    (1, '2020-01-01', 15),
    (2, '2020-01-02', 10),
    (3, '2020-01-03', 5);

仔:

select ID, Date, Amount, 
    sum(Amount) over (order by Date) as RunningBalanceForward, 
    sum(Amount) over (order by Date desc) as RunningBalanceBackwards
from @sales
order by ID

结果：

除了总运行，使用一个row_number创建的排序列和分区通过id以便通过日期desc产生rowid然后便通过rowid

只是为了显示一个例子是该分区使用。

declare @Sales table (
  ID int identity(1,1) primary key,
  [Date] date,
  Store varchar(30),
  Amount int
  
);

insert into @Sales ([Date], Store, Amount)
values
  ('2020-01-01','A',1), ('2020-01-03','A',1)
, ('2020-01-05','A',1)
, ('2020-01-02','B',10), ('2020-01-04','B',10)
, ('2020-01-06','B',10)
;
    
select Store, [Date], Amount, 
    sum(Amount) over (partition by Store order by [Date] ASC) as RunningTotal, 
    sum(Amount) over (partition by Store order by [Date] DESC) as ReverseRunningTotal
from @Sales
order by Store, [Date] 
GO

储存|日期|数量|RunningTotal|ReverseRunningTotal
:---- | :--------- | -----: | -----------: | ------------------:
一个| 2020-01-01 | 1 | 1 | 3
一个| 2020-01-03 | 1 | 2 | 2
一个| 2020-01-05 | 1 | 3 | 1
B| 2020-01-02 | 10 | 10 | 30
B| 2020-01-04 | 10 | 20 | 20
B| 2020-01-06 | 10 | 30 | 10