Re: how to improve impdp

From: Charles Schultz <sacrophyte_at_gmail.com>
Date: Tue, 9 Jul 2013 09:36:44 -0500
Message-ID: <CAPZQniXTAj1P5F8AT_pjwE86+MSp7qm0RqHbiR6DNya9KnOVpg_at_mail.gmail.com>

+1 - this is very similar to what we use when we need to use datadump on our large databases. Tried to convince Roy Swonger and Carol Palmer to enhance indexes/constraints/stats back in 2007, but I guess it never hit the Top 10 list. Too bad.

On Tue, Jul 9, 2013 at 9:25 AM, rjamya <rjamya_at_gmail.com> wrote:

> You have only one dmp file so parallel isn't much use here. I bet most of
> your time is going in creating indexes and enabling constraints.
> I'd expdp with
> dumpfile=filename.%u.dmp parallel=5
>
> and then use similar params for impdp. That will ONLY help your with table
> data import. Unless your tables have degree > 1 (in source db) constraints
> will be enabled in non-parallel mode. Unless your indexes have degree > 1
> (in source db) they will be created by a single process, i.e. parallel will
> be used by 4 DW threads will be idling..
>

-- 
Charles Schultz


--
http://www.freelists.org/webpage/oracle-l

Received on Tue Jul 09 2013 - 16:36:44 CEST