Walter,

It's got nothing to do with the tool ... it's got everything to do with client/server processing.

Jojo was talking about pulling down all his data to the client (his Form) and that's something you don't need to do. What you're talking about with your statistical processing is something that should be done server-side, not client-side.

And when I say "server-side", I don't necessarily mean in SQL Server itself (although with SQL2005's capability of being able to use managed code now, you can probably write more powerful stuff than you're able to with T-SQL ... I haven't experimented with this aspect yet).

Client/server apps should have server-side business classes where you can do datamunging if you need to. Since it's done server-side, those vast amounts of data do not need to travel down the wire to the client-side.

~~Bonnie



>Bonnie,
>
>>I don't know what your application does, but how on earth can one Form use or process 21,000 records? You don't need to retrieve the entire table into one DataSet. You pull down what you need, as you need it. Here's an example:
>
>What about statistical processing ?? I'm stunned again and again that pulling out large amounts of data to the client is bad per se. Sure, if you can avoid it, you should. However some types of applications just ask for it. If I want to make a statistical analysis I might HAVE to download all data before post processing.
>
>I find too often the argument is made that its bad because the tool cannot handle that comfortably. IOW the argument is influcence by the limitations of the tool, not of what from an ideal perspective should be best.
>
>Walter