最近忙于公司BI軟件性能測試，這幾天主要測試CUBE采用ROLAP下，PA的并發和穩定性。涉及表和維度，立方：事實表sales_fact_1997,維度表time_by_day；立方:sales。修改內容：刪除原來的TIME維度，新建TIME維度，修改SALES立方。

數據插入：

1、事實數據插入：通過DTS加調度實現將sales_fact_1997的數據進行復制。頻率為每分鐘10000條。主要用到的SQL語句：select top 10000 * from sales_fact_1997

2、維度數據插入：通過SQL語句插入數據到time_by_day.

測試用到的SQL語句：

1、單條插入

INSERT INTO time_by_day

(time_id, the_date, the_year, month_of_year, quarter,day_of_month)

VALUES ('1101', '1999-10-1', '1999', '10', 'Q4','1')
 

2、單條插入:

INSERT INTO time_by_day

(time_id, the_date, the_year, month_of_year, quarter, day_of_month)

SELECT TOP 1 time_id + 1 AS time_id, the_date + 1 AS the_date, YEAR(the_date + 1)

AS the_year, MONTH(the_date + 1) AS month_of_year, { fn QUARTER(the_date + 1)

} AS quarter, DAY(the_date + 1) AS day_of_month

FROM time_by_day

ORDER BY time_id DESC
 

3、循環插入：

DECLARE @MyCounter INT

SET @MyCounter = 0            /*設置變量*/

WHILE (@MyCounter < 2)     /*設置循環次數*/

BEGIN

WAITFOR DELAY '000:00:10'   /*延遲時間10秒*/

INSERT INTO time_by_day
(time_id, the_date, the_year, month_of_year, quarter, day_of_month)

SELECT TOP 1 time_id + 1 AS time_id, the_date + 1 AS the_date, YEAR(the_date + 1)

AS the_year, MONTH(the_date + 1) AS month_of_year, { fn QUARTER(the_date + 1)

} AS quarter, DAY(the_date + 1) AS day_of_month

FROM time_by_day

ORDER BY time_id DESC

SET @MyCounter = @MyCounter + 1

END
 

4、插入以時間為變量的數據

DECLARE @MyCounter INT

declare @the_date datetime

SET @MyCounter = 0

SET @the_date = '1999-1-4'

WHILE (@MyCounter < 200000)

BEGIN

WAITFOR DELAY '000:00:10'

/*INSERT INTO time_by_day

(time_id, the_date, the_year, month_of_year, quarter, day_of_month)

SELECT TOP 1 time_id + 1 AS time_id, the_date + 1 AS the_date, YEAR(the_date + 1)

AS the_year, MONTH(the_date + 1) AS month_of_year, { fn QUARTER(the_date + 1)

} AS quarter, DAY(the_date + 1) AS day_of_month

FROM time_by_day

ORDER BY time_id DESC

*/

insert into time_by_day (time_id,the_date)values('371',@the_date)

SET @the_date = @the_date + 1

SET @MyCounter = @MyCounter + 1

END