Re: Why would an Insert /*+ append */ be slower than a straight insert?

"Jusung Yang" <JusungYang_at_yahoo.com> wrote in message news:130ba93a.0305231805.2ff2c3ea_at_posting.google.com...
> Just a thought. You may not be comparing regular insert with direct
> insert appropriately. You have the two different tables on two
> different tablespaces, yes? The right way to do it would be to use
> the same table for both regular and direct insert and compare the
> results. Whow knows what differences there are between the two tables
> on the physical level. You can dig into it if you wish, but that's not
> the right way to do any comparison, IMO.
>
>
> - Jusung Yang

my bad on the explanation. They are the same tables for each test. I use a staging table in one tablespace and a master table in another tablespace. I truncate the master table between loads, take the tablespaces offline(to flush the buffer cache), and flush the shared pool.

exact duplicates. all that changes in the code is that I add append. I read that append inserts above the highwater mark... Well we are using very small extents plus we keep our datafile only slightly larger than our highwater mark, Im willing to bet the slow down is due to the autoextends. We need to keep the datafile small since we have to send it across the network every day and this is the slowest part of the load process.

I found a better way to do this. Im going to sql load directly into my master table. My .badfile will be then sqlloaded into my stage table. Those will be the records I need to update and delete. Thereby totally bypassing the need to insert. Received on Sat May 24 2003 - 15:11:10 CDT