Re: How to improve performance on dataload process

  • From: Mladen Gogala <gogala@xxxxxxxxxxxxx>
  • To: roon987@xxxxxxxxxxx
  • Date: Wed, 28 Sep 2005 11:56:52 +0000

On 09/28/2005 05:54:26 AM, Zabair Ahmed wrote:
> 9.2.0.6 on Solaris 9
>  
> Am loading data into a staging database using a ETL tool (Informatica) 
> and have done the following to improve the performance of the dataload 
> process
>  
> 1) Database in no-archive log mode
> 2) Indexes  being built after the dataload
> 3) Load being done in parallel
> 4) Nologging being used
> 5) Tables not being analysed during the dataload process
>  
> What else can I do to improve the performance of this load process?
>  

First of all, find out how active your database is and how efficient is your 
tool.
Are you spending time in the database or elsewhere? 
Your tool might be, for all described here, transferring row by row over the 
network,
not using array interface and thus increasing the number of packets by an order 
of magnitude,
slowing the whole thing down. Perl was infamous for doing that as it wasn't 
supporting 
array interface until very recently.

-- 
Mladen Gogala
http://www.mgogala.com


--
//www.freelists.org/webpage/oracle-l

Other related posts: