I have a rather large table of 50000 records, and I want to cut this down to 5000. How would I write an SQL query to delete the other 45000 records. The basic table structure contains the column of a datetime.
A rough idea of the query I want is the following
DELETE FROM mytable WHERE countexceeded(5000) ORDER BY filedate DESC;
I could write this in C# somehow grabbing the row index number and doing some work around that, however is there a tidy way to do 开发者_开发问答this?
The answer you have accepted is not valid syntax as DELETE
does not allow an ORDER BY
clause. You can use
;WITH T AS
(
SELECT TOP 45000 *
FROM mytable
ORDER BY filedate
)
DELETE FROM T
DELETE TOP(45000) FROM mytable ORDER BY filedate ASC;
Change the order by to ascending to get the rows in reverse order and then delete the top 45000.
Hope this helps.
Edit:-
I apologize for the invalid syntax. Here is my second attempt.
DELETE FROM myTable a INNER JOIN
(SELECT TOP(45000) * FROM myTable ORDER BY fileDate ASC) b ON a.id = b.id
If you do not have a unique column then please use Martin Smith's CTE answer.
if the table is correctly ordered:
DELETE FROM mytable LIMIT 5000
if not and the table has correctly ordered auto_increment index:
get the row
SELECT id, filedate FROM mytable LIMIT 1, 50000;
save the id and then delete
DELETE FROM mytable WHERE id >= @id;
if not ordered correctly, you could use filedate instead of id, but if it's a date without time, you could get undesired rows deleted from the same date, so be carefull with filedate deletion solution
精彩评论