Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Direct Path Export vs. Conventional Path Export

Re: Direct Path Export vs. Conventional Path Export

From: Anurag Varma <avarmadba.skipthis_at_yahoo.com>
Date: Tue, 02 Sep 2003 20:28:05 GMT
Message-ID: <p375b.18504$zL5.12749@nwrdny02.gnilink.net>

"Goran D." <goran99_makni_at_yahoo.com> wrote in message news:bj1utr$c0c$1_at_fegnews.vip.hr...
>
> "Burton Peltier" <burttemp1REMOVE_THIS_at_bellsouth.net> wrote in message
> news:8jV4b.4779$UF.3526_at_bignews2.bellsouth.net...
> > I recently switched our export backups to direct path and was impressed
at
> > the performance boost.
> >
> > I did not set the recordlength (or buffer before with conventional path)
> and
> > still am seeing a significant boost in performance. The docs mention
these
> > settings to boost performance.
> >
> > In every case, the direct path is at least twice as fast and sometimes 3
> > times faster than the conventional path.
> >
> > I assume I am missing something here because this seems too good to be
> true,
> > so I looked up this in the docs.
> >
> > I don't see any disadvantage other than you must use the same character
> set
> > for the client exp session as the database character set, which I always
> do
> > anyway.
> >
> > Anyone have experience with this and can comment good or bad on direct
> path?
> >
>
> There is a serious bug; certain versions (817X, 9XXX?) on certain
platforms
> create a corrupt dump file - that cannot be imported - if a table is
> exported via direct path that contains either chained or migrated rows.
This
> is easily reproducible and should be tested before using direct path
export.
>
> Regards...
>
>

.... there are serious bugs with the non-direct path also. So yes, OP should test any method he implements.

Anurag Received on Tue Sep 02 2003 - 15:28:05 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US