Re: Importing huge dump file.

On Sep 7, 7:44 am, "fitzjarr..._at_cox.net" <fitzjarr..._at_cox.net> wrote:
> On Sep 7, 2:33 am, aina.t..._at_gmail.com wrote:
>
> The buffer and recordlength parameters are, essentially, mutually
> exclusive. buffer is used by conventional path imports, where
> recordlength is used by direct path imports. Yes, when exporting/
> importing LOB data a direct path export/import will use conventional
> path, and it may be necessary to specify both parameters in the
> parameter list. No errors are thrown when both are specified, but
> know that it is one or the other which is used, depending upon the
> mode of the export/import. And what good is your 'advice' to 'Put DMP
> file on other disk than the tablespace file'? I don't know of any DBA
> who would knowingly and willingly place a dmp file on a filesystem
> dedicated to datafiles/tempfiles. And the OP may have space
> considerations you know nothing about.
>

I do it all the time. SAME it's all the, you know.

Oh wait, I'm not a DBA. Never mind. :-)

I do get 26G dmp into a partial overnight window, though. commit=y buffer=20000000 statistics=recalculate, no significant LOBS. Not even sure why I don't use recordlength, I do in the exports. Prolly some old bug set me off it. ...there is no DIRECT parameter for import? recordlength on imp only has to do with different platforms? I hate being confused about something I've used for years. Oh wait, I've seen this before - it's SQL*Loader that has the direct load option, not imp. Right? Or have I totally confused myself?

jg

--
@home.com is bogus.
Don't shoot milk out of your nose:  http://www.npr.org/templates/story/story.php?storyId=14204620