Question

Working environment is SQL Server 2000. I have a table with no indexes, no PK… Total number of rows is 600,000.

How can I update a column from row 0 -> 100,000 with a value then from 100,001 -> 200,000 with another, and so on?

Thank you.

Answer 1

set all values to null, then SET ROWCOUNT 100000. Then do successive updates ( each will affect 100000 rows) with different values where that column IS NULL. @@rowcount will be the number of rows affected after each update, so stop when it is less than 100000.

For @Shannon’s comment, the ROWCOUNT will not be honored for update/delete/insert statements in the next version of SQL Server (post-SQL Server 2008), but it will work fine for SQL Server 2000. The recommended change is to use the TOP clause, but I don’t think that is supported for updates until SQL Server 2005.

I think you could sue a cursor if you wanted…

update mytable set myid = null

SET NOCOUNT ON -- prevent all those "1 row(s) updated" messages
declare @count int, @value int, @myid int
set @count = 1
set @value = 1
declare cursor_update cursor for select myid from mytable
open cursor_update
fetch cursor_update into @myid
while @@FETCH_STATUS = 0
begin
    update mytable set myid = @value where current of cursor_update

    set @count = @count + 1
    if (@count > 100000) 
    begin
        set @count = 1
        set @value = @value + 1
    end

    fetch cursor_update into @myid
end
close cursor_update
deallocate cursor_update