RE: Multi-row insert in SQL

[Vern Hamberg <vhamberg@xxxxxxxxxxxxxxxxxxxxxxxxx>]

Yeah, I snuck iSQL inthere, didn't I?

But the thing thatfinally triggered by post was an article on improving 
performance in stored procedures - aimed at the non-iSeries versions, but 
good info anyway.

In there they say to use one statement instead of many - it'll be faster, 
as each statement has some overhead. This perhaps comes into play more with 
things where you need to install master files or test files. Now, I think 
that, e;g;, TPC-H does individual inserts, partly because they calculate 
the values for each row, and because the syntax does not exist in all 
implementations of SQL.

Anyway, kind of an interesting discussion. Thanks

Vern

At 09:04 PM 6/13/2003 -0400, you wrote:
> midrange-l-request@xxxxxxxxxxxx wrote:
>
> >   5. RE: Multi-row insert in SQL (Vern Hamberg)
> >
> >Right, Tom. But, of course, you need to have populated the intermediate
> >table somehow.  ;-)
> >
> >It seems that the mechanism for adding multiple records exists in the
> >engine, since both the MODS RPG process and the subselect process are 
> handled.
> >
> >Right now, the place I'd find the unsupported syntax to be helpful is in
> >interactive SQL, when I'm trying to populate a file for testing.
>
> Sorry, Vern, I should've seen ISQL coming. For me, maybe the choice would 
> be to use EDTF to type into a CSV file and then CPYFRMIMPF to get them 
> into a database format. I know, that might as well put them into your 
> final table; but maybe you need to append sets of rows to your table or 
> massage them otherwise before INSERT. And THEN the subselect...?
>
> Tom Liotta