If you drop from text to varchar 500, any record that contains 501+ characters will be truncated to 500. That's where you would have a loss of data.

Nice to see you back from whatever you were doing .

WOw, thanx for the complete details... can you tell us which query is failing and why you think it's not working?

Also some sample data and expected output would be usefull.

Rewrite the report so that he gets less data than the whole db...

What is the report about that it requires downloading so much data?

Can you give us more details about the task at hand... seems to me like you need to use a trigger to achieve this operation.

Why do you want to do that?

Why does he do that?

I'm not sure but that can't help the server... in any way. Maybe that would be enough of a boost to save the day, but I doubt it.

"The additional column to be populated in table2 is a datetime with a default value."

Not set based... what are you gonna do if there's 1 M names in there??

Here's an exemple :

SELECT O.XType

, O.name

FROM dbo.SysObjects O

WHERE ID IN (SELECT TOP 90 PERCENT ID FROM dbo.SysObjects O2 WHERE O2.XType = O.XType order by O2.Name)

ORDER BY O.XType, O.Name

Here's an exemple on a big table WITH timestamp :

exec sp_executesql N'UPDATE "Ideal"."dbo"."Chaudières" SET "Succursale"=@P1 WHERE "no"=@P2 AND "TsAccess"=@P3', N'@P1 nvarchar(1),@P2 int,@P3 varbinary(8)', NULL, 867, 0x0000000000102078

Here's the SAME UPDATE...

With a good progress meter .

The timestamp is a column maintained by the server. Everytime a row is updated the timestamp is modified by the server (like a row version thing). So intead...

I think he needs to keep the table under 1000 rows. I already used that technic before and it works well. An instead of insert trigger can also...