http://sqlfiddle.com/#!18/f8707c/2

CREATE TABLE Emp ([Num] varchar(2)) ;
INSERT INTO Emp ([Num]) VALUES ('AA'), ('BB'), ('CC'), ('DD') ;

CREATE TABLE Center ([ID] varchar(2), [EmpQty] int) ;
INSERT INTO Center ([ID], [EmpQty]) VALUES ('01', 2), ('02', 3), ('03', 2) ;

CREATE TABLE WorkDate ([SchDate] varchar(10)) ;
INSERT INTO WorkDate ([SchDate]) VALUES ('2018-07-27') ;
INSERT INTO WorkDate ([SchDate]) VALUES ('2018-07-28') ;
INSERT INTO WorkDate ([SchDate]) VALUES ('2018-07-29') ;

select d.Row1,d.SchDate,d.ID,e.Num
from (
select ROW_NUMBER() OVER(ORDER BY f.SchDate,c.ID) AS Row1,f.SchDate,c.ID,c.EmpQty
from WorkDate as f ,(
select b.ID,b.EmpQty
from Center as a
Cross Apply Center as b
where 1=1
and a.EmpQty<=b.EmpQty
) as c
) as d
left join
(select (ROW_NUMBER() OVER(ORDER BY Num asc) %4 ) AS Row2,Num from Emp) as e
on d.Row1 % 4 = e.Row2

我的解法也是寫一個SP，邏輯寫在註解裡面
有想到優化方式，再來優化 :D

db<>fiddle 線上測試DEMO

TEST DDL:

CREATE TABLE Emp ([ID] varchar(2)) ; INSERT INTO Emp ([ID]) VALUES ('AA'), ('BB'), ('CC'), ('DD') ; 

CREATE TABLE Center ([ID] int, [EmpQty] int) ; INSERT INTO Center ([ID], [EmpQty]) VALUES (01, 2), (02, 3), (03, 2) ; 

CREATE TABLE Sch ([Datetime] datetime, [CenterID] int, [EmpID] varchar(2)) ;

SP:

CREATE PROCEDURE SP_SCH_Center_ByEmp   
    @v_date datetime  
AS  

declare @v_firstEmp nvarchar(2) =  (
	--找出昨天誰排班最少，藉由group + rownumber count,接著篩選出rank = 1的資料
	select [EmpID] 
	from (
		select [EmpID]
			,row_number() over (order by count(1),[EmpID]) rnk 
		from Sch
		where [Datetime] = (@v_date-1)
		group by [EmpID]
	) T where rnk = 1 
);
declare @v_empCount int = (select count(1) from Emp);

--如果@v_firstEmp沒資料，以第一個員工為準
--如果@v_firstEmp有資料，以前天排最少班的第一個員工為準
if @v_firstEmp is null begin
	select @v_firstEmp = (select top 1 [ID] from Emp);
end;

--先把當天Center資料展出來(使用遞迴)
WITH CTE AS (
	SELECT 1 as Number,ID,EmpQty from Center
	UNION ALL
	SELECT Number+1,ID,EmpQty
	FROM CTE 
	WHERE Number < EmpQty 
)
--幫員工做一個排序
,CTE2 as (
	select * , row_number() over (order by [ID] ) rnk from Emp
)
--Center資料做一個排序，特別使用站點排名 + @v_firstEmp員工的排名-2 做%的基準數
,CTE3 as (
	SELECT @v_date [Datetime],ID [CenterID]
		,( row_number() over (order by [ID]) 
			+ (select rnk - 2 from CTE2 where [ID] = @v_firstEmp )
		) rnk
	FROM CTE
)
insert into sch 
	select T1.[Datetime],T1.[CenterID],T2.[ID] 
	from CTE3 T1
	left join CTE2 T2 on ((T1.rnk) % (@v_empCount))  = T2.rnk-1;

測試新增7/27,28的資料

exec SP_SCH_Center_ByEmp '2018-07-27';
exec SP_SCH_Center_ByEmp '2018-07-28';

select * from SCH;

得到結果:

另外，這間公司這麼血汗，一個人排這麼多點站一天 XDD

create table ithelp180727_emp (
    id serial primary key
,   ename text not null
);

insert into ithelp180727_emp (ename) values
('AA'), ('BB'), ('CC'), ('DD');

create table ithelp180727_stat (
    sid serial primary key
,   matching int not null
);

insert into ithelp180727_stat values
(1, 2), (2, 3), (3, 2);

--
with   st1 as (
select max(matching) mm
  from ithelp180727_stat
), st2 as (
select count(*) as cnt
  from ithelp180727_emp
), st3 as (
select date '2018-07-27' + d as dt
     , sid
  from st1
     , generate_series(1, st1.mm) as i
     , generate_series(1, i) as j
     , generate_series(0, 1) as d
     , ithelp180727_stat s
 where s.matching = i
), st4 as (
select *
     , row_number() over(order by dt, sid) rn
  from st3
), st5 as (
select dt, sid, rn
     , ntile(st2.cnt::integer) over(order by mod(rn::integer - 1, st2.cnt::integer)) as asgnemp
  from st4
     , st2
)
select dt, sid, ename
  from st5
     , ithelp180727_emp e
 where st5.asgnemp = e.id
order by dt, sid, rn
;

     dt     | sid | ename 
------------+-----+-------
 2018-07-27 |   1 | AA
 2018-07-27 |   1 | BB
 2018-07-27 |   2 | CC
 2018-07-27 |   2 | DD
 2018-07-27 |   2 | AA
 2018-07-27 |   3 | BB
 2018-07-27 |   3 | CC
 2018-07-28 |   1 | DD
 2018-07-28 |   1 | AA
 2018-07-28 |   2 | BB
 2018-07-28 |   2 | CC
 2018-07-28 |   2 | DD
 2018-07-28 |   3 | AA
 2018-07-28 |   3 | BB
(14 筆資料列)

----------------------

insert into ithelp180727_stat values (4,1);
-- 指令一樣
with   st1 as (
select max(matching) mm
  from ithelp180727_stat
), st2 as (
select count(*) as cnt
  from ithelp180727_emp
), st3 as (
select date '2018-07-27' + d as dt
     , sid
  from st1
     , generate_series(1, st1.mm) as i
     , generate_series(1, i) as j
     , generate_series(0, 1) as d
     , ithelp180727_stat s
 where s.matching = i
), st4 as (
select *
     , row_number() over(order by dt, sid) rn
  from st3
), st5 as (
select dt, sid, rn
     , ntile(st2.cnt::integer) over(order by mod(rn::integer - 1, st2.cnt::integer)) as asgnemp
  from st4
     , st2
)
select dt, sid, ename
  from st5
     , ithelp180727_emp e
 where st5.asgnemp = e.id
order by dt, sid, rn
;

     dt     | sid | ename 
------------+-----+-------
 2018-07-27 |   1 | AA
 2018-07-27 |   1 | BB
 2018-07-27 |   2 | CC
 2018-07-27 |   2 | DD
 2018-07-27 |   2 | AA
 2018-07-27 |   3 | BB
 2018-07-27 |   3 | CC
 2018-07-27 |   4 | DD
 2018-07-28 |   1 | AA
 2018-07-28 |   1 | BB
 2018-07-28 |   2 | CC
 2018-07-28 |   2 | DD
 2018-07-28 |   2 | AA
 2018-07-28 |   3 | BB
 2018-07-28 |   3 | CC
 2018-07-28 |   4 | DD
(16 筆資料列)