Question

I’m trying to build a table in MS SQL 2008, loaded with roughly 50,000 rows of data. Right now I’m doing something like:

Create Table MyCustomData
(
ColumnKey Int Null,
Column1 NVarChar(100) Null,
Column2 NVarChar(100) Null

Primary Key Clustered
(
ColumnKey ASC
)
WITH (
PAD_INDEX  = OFF, 
STATISTICS_NORECOMPUTE  = OFF, 
IGNORE_DUP_KEY = OFF, 
ALLOW_ROW_LOCKS  = ON, 
ALLOW_PAGE_LOCKS  = ON
)
)

CREATE INDEX IDX_COLUMN1 ON MyCustomData([COLUMN1])
CREATE INDEX IDX_COLUMN2 ON MyCustomData([COLUMN2])

DECLARE @MyCount Int

SET @MyCount = 0

WHILE @MyCount < 50000
BEGIN
 INSERT INTO MyCustomData
 (ColumnKey, Column1, Column2)
 Select @MyCount + 1, 'Custom Data 1', 'Custom Data 2'

Set @MyCount = @MyCount + 1
END

My problem is that this is crazy slow. I thought at one point I could create a Select Statement to build my custom data and use that as the datasource for my Insert Into statement.

i.e. something like

INSERT INTO MyCustomData
 (ColumnKey, Column1, Column2)
From (Select Top 50000 Row_Count(), 'Custom Data 1', 'Custom Data 2')

I know this doesn’t work, but its the only thing I can show that seems to provide an example of what I’m after. Any suggestions would be greatly appriciated.

Answer 1

Well, if I wanted 50000 rows with constant data and a counter, I’d go

select row_number() over(order by t1.[number]), 'Custom data 1', 'Custom data 2'
from
  master..spt_values as t1
  cross join (select [number] from master..spt_values where [type] = 'P' and [number] between 1 and 50) as t2
where
  [type] = 'P' and t1.[number] between 1 and 1000