update rows if different

[webnick]

What is the best way to UPDATE a row only if values are different? We have a mainframe extract that literally has 100,000 rows and am worried about the performance of just running INSERTs each night.

Is there a simple comparison command or would you have to nest a SELECT statement? Thank you,

Nick

[tomhath]

Can you give some more details?

I think you're asking if it's better to update the existing data versus generate a complete new table every night. Then answer of course is: "It depends".

How many rows change? 100's 1000's?

Loading the data from a csv file shouldn't take very long. Have you timed it?