Goran wrote:
> "Mark Bole" <makbo_at_pacbell.net> wrote in message
> news:exovd.33590$zx1.24975_at_newssvr13.news.prodigy.com...
>
>>Goran wrote:
>>
>>>"zeb" <spam_at_nowhere.com> wrote in message
>>>news:41bdfca9$0$11630$636a15ce_at_news.free.fr...
>>>
>>>
>>>>Hi,
>>>>
>>>>It seems direct path for export and SQL Loader is always
>>>>quicker than conventional path ...
>>>>
>>>>What are the drawbacks for direct path ?
>>>>Why not to use always direct path ?
>>>>When it is better to use diret path and
>>>>when it is better to use conventional path ?
>>>>
>>>>Thanks for your lights ...
>>>>
>>>>Oracle 8i, 9i
>>>>HP-UX 11.0 AIX 5L
>>>>
>>>
>>>
>>>Direct path export (8i, not sure about other versions, not sure if it's
>>>platform specific or generic) behaves very wrong if a table has
>>>chained/migrated rows. So wrong that it produces unusable dump. Easy to
>>>reproduce...
>>>
>>>Regards...
>>>
>>>
>>
>>It's an old version, but just as a matter of curiousity, do you remember
>>what specific imp/exp error(s) indicated that the export file was
>>unusable? I supported a system that did daily direct path exports and
>>although re-importing them (into another database) was rare, we never
>>had any problems. Never checked for presence of chained/migrated rows...
>>
>>-Mark Bole
>>
>
>
> Doc ID# 2690640
> Regards..
>
>
EXP from 8.1.7.2 to 8.1.7.4 could well have been subject to bugs. I'll
take your word for it, as I'm not in a position to test one way or the
other.
My main concern in asking the question was to differentiate between
direct path export and direct path SQL*Loader. The former in my
experience has been pretty clean, simple, and advantageous.
Mark Bole
Received on Mon Dec 13 2004 - 20:35:16 CST