Re: Direct Path Export vs. Conventional Path Export

"Anurag Varma" <avdbi_at_hotmail.com> wrote in message news:<nKf5b.1194$uh6.956_at_news02.roc.ny>...
> "Burton Peltier" <burttemp1REMOVE_THIS_at_bellsouth.net> wrote in message news:Jte5b.1410$Z5.296_at_bignews2.bellsouth.net...
> > After reading the replies to my original post, I am more concerned about
> > BOTH direct and conventional exports.
> >
> > Although both have their share of bugs, I have used the conventional path
> > for so long, I will stick with it.
> >
> > FYI... an interesting article in Metalink is at: Note:155477.1 "Parameter
> > DIRECT: Conventional Path Export Versus Direct Path Export"
> >
>
> :) You've stuck with conventional export .. does not mean that you will never hit a bug in future versions.
> In most versions ... Direct path export is not any worse or better than conventional export in terms of bugs.
> You can easily test out if you are hitting a bug ... just export import .. and compare data
> in the two instances.

This is very wrong. The whole problem with bugs in exp/imp isn't that you can compare data - it's that you might hit the bug in the future when it is too late.

Especially in the recent past, where merely doing a point upgrade can break it, and you _have_ to do the upgrade to be supported.

Note also one of the bugs was in buffer size...

To the OP: if you use direct on imp, be sure you understand all ramifications.
It can work well, but when something messes up (like someone interrupting a big load), it can be all-nighter time.

>
> Of course if there is no need to speed up the export .. then why bother.
> .. and if you are critically relying on export as a backup .. then re-check your conventional option.
> How do you know you are not hitting a bug already :)

:-) It's not like Oracle figures it out before customers get it!

jg

--
@home.com is bogus.
http://www.newsfilter.org/oddarchive/index.html