Re: 11gR2 Data Pump question - parallel setting

From: rjamya <rjamya_at_gmail.com>
Date: Fri, 27 Jul 2012 09:47:10 -0400
Message-ID: <CAGurbTNTe3gAS=d2wN8Ma1-jOyqCDtZ=+iQ8ozyh-i5o2TRprA_at_mail.gmail.com>

I am sure they will catch up on these suggestions, eventually.

Case in point, back in 2003, I wanted the ability to have dbms_stats spawn parallel threads to collect stats to minimize collection time, there wasn't any. So, I ended up writing my own plsql and jobs based solution (ref: http://www.orafaq.com/maillist/oracle-l/2003/03/06/0415.htm). Now that functionality is available in recent versions (https://blogs.oracle.com/optimizer/entry/gathering_optimizer_statistics_is_one), although I have not yet tested it.

In my previous job, we ran into exact problem that Jon had, so a combination of (include= and sqlfile=) options for impdp followed by a bit of sed magic helped in speeding up the process. Alas the previous job wouldn't let me take out a single script (even generic one) with me. And I am pretty sure those are still in my home directory on unix :)

Raj

On Fri, Jul 27, 2012 at 9:06 AM, Charles Schultz <sacrophyte_at_gmail.com> wrote:
> I am very curious, what reason does Oracle give for not incorporating Greg's and Raj's 'solution'? Is there a good reason why those DDL statements cannot truly parallelized in multiple worker processes within datapump?
>
> References: http://www.freelists.org/post/oracle-l/Script-to-build-indexes-in-parallel,5

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

Received on Fri Jul 27 2012 - 08:47:10 CDT