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: Data Pump slower than equivalent Export ?!?

Re: Data Pump slower than equivalent Export ?!?

From: Ronald Rood <devnull_at_ronr.nl>
Date: Fri, 16 Sep 2005 20:18:05 +0200
Message-ID: <0001HW.BF50D8FD021EBD40F0305550@news.individual.net>


On Fri, 16 Sep 2005 16:19:29 +0200, mccmx_at_hotmail.com wrote (in article <1126880369.250882.129120_at_z14g2000cwz.googlegroups.com>):

> Oracle 10.1.0.4 EE on RHEL 3
>
> Oracle claim that Data Pump is a considerably faster method of loading
> and unloading data to/from the database.
<snip>
> Export takes 43 seconds and Data Pump takes over 3 minutes..!
>
> 95% of the delay in Data Pump appears to be in the step:
>
> "Estimate in progress using BLOCKS method..."
>
> What exactly is done during this step..?

Hi Matt,
I am quite sure there can be made some improvements in the data pump but for a serious speed test you need to take a larger database. I just finished some testing and was quite happy with it. impdp read 46GB of data from a source database and stored in a target database in less than 2 hours, over a network link (100Mb). The ftp session alone of the conventional expdat.dmp (49GB) of the same database took 1 hour and 20 minutes... Disappointing is the fact that only for the data impdp uses parallel workers and not for the other objects. They are serialized and in my case that took nearly 10 hours to complete. Conventional exp (full) took 10 hours to complete.
So, I think impdp deserves some more testing ...

Completely missing are warnings about NLS problems. Conventional exp did not much but it did give warnings ...

With kind regards / met vriendelijke groeten, Ronald

http://ciber.nl
http://homepage.mac.com/ik_zelf/oracle Received on Fri Sep 16 2005 - 13:18:05 CDT

Original text of this message

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