How to delete the top 1000 rows from a table using Sql Server 2008?

SqlSql ServerSql Delete

Sql Problem Overview


I have a table in SQL Server. I would like to delete the top 1000 rows from it. However, I tried this, but I instead of just deleting the top 1000 rows it deleted all the rows in the table.

Here is the code:

delete from [mytab] 
select top 1000 
a1,a2,a3
from [mytab]

Sql Solutions


Solution 1 - Sql

The code you tried is in fact two statements. A DELETE followed by a SELECT.

You don't define TOP as ordered by what.

For a specific ordering criteria deleting from a CTE or similar table expression is the most efficient way.

;WITH CTE AS
(
SELECT TOP 1000 *
FROM [mytab]
ORDER BY a1
)
DELETE FROM CTE

Solution 2 - Sql

May be better for sql2005+ to use:

DELETE TOP (1000)
FROM [MyTab]
WHERE YourConditions

For Sql2000:

DELETE FROM [MyTab]
WHERE YourIdField IN 
(
  SELECT TOP 1000 
    YourIdField 
  FROM [MyTab]
  WHERE YourConditions
)

BUT

If you want to delete specific subset of rows instead of arbitrary subset, you should explicitly specify order to subquery:

DELETE FROM [MyTab]
WHERE YourIdField IN 
(
  SELECT TOP 1000 
    YourIdField 
  FROM [MyTab]
  WHERE YourConditions
  ORDER BY ExplicitSortOrder
)

Thanks tp @gbn for mentioning and demanding the more clear and exact answer.

Solution 3 - Sql

As defined in the link below, you can delete in a straight forward manner

USE AdventureWorks2008R2;
GO
DELETE TOP (20) 
FROM Purchasing.PurchaseOrderDetail
WHERE DueDate < '20020701';
GO

http://technet.microsoft.com/en-us/library/ms175486(v=sql.105).aspx

Solution 4 - Sql

delete from [mytab]
where [mytab].primarykeyid in
(
select top 1000 primarykeyid
from [mytab]
)

Solution 5 - Sql

SET ROWCOUNT 1000;

DELETE FROM [MyTable] WHERE .....

Solution 6 - Sql

It is fast. Try it:

DELETE FROM YourTABLE
FROM (SELECT TOP XX PK FROM YourTABLE) tbl
WHERE YourTABLE.PK = tbl.PK

Replace YourTABLE by table name, XX by a number, for example 1000, pk is the name of the primary key field of your table.

Solution 7 - Sql

I agree with the Hamed elahi and Glorfindel.

My suggestion to add is you can delete and update using aliases

/* 
  given a table bi_customer_actions
  with a field bca_delete_flag of tinyint or bit
    and a field bca_add_date of datetime

  note: the *if 1=1* structure allows me to fold them and turn them on and off
 */
declare
		@Nrows int = 1000

if 1=1 /* testing the inner select */
begin
  select top (@Nrows) * 
    from bi_customer_actions
    where bca_delete_flag = 1
    order by bca_add_date
end

if 1=1 /* delete or update or select */
begin
  --select bca.*
  --update bca  set bca_delete_flag = 0
  delete bca
    from (
      select top (@Nrows) * 
        from bi_customer_actions
        where bca_delete_flag = 1
        order by bca_add_date
    ) as bca
end 

Attributions

All content for this solution is sourced from the original question on Stackoverflow.

The content on this page is licensed under the Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) license.

Content TypeOriginal AuthorOriginal Content on Stackoverflow
QuestionedgarmtzeView Question on Stackoverflow
Solution 1 - SqlMartin SmithView Answer on Stackoverflow
Solution 2 - SqlOleg DokView Answer on Stackoverflow
Solution 3 - SqlRishi BhatnagarView Answer on Stackoverflow
Solution 4 - SqlJason DamView Answer on Stackoverflow
Solution 5 - SqlJoe BourneView Answer on Stackoverflow
Solution 6 - SqlHamed elahiView Answer on Stackoverflow
Solution 7 - SqlyeOldeDataSmytheView Answer on Stackoverflow