Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: Rollback segments
Hi,
just a little info here.
You can look at the problem in several ways. More than one way to eat a hard
boiled egg.
1. Select/Insert with out rollbacks yes this is possible.
You can perform even parallel select's where you can determine how many
processes you wish to use in function with how many CPU's you have. bla bla
The aim would be to use SQL Loader with direct path and unrecoverable options.
When you extract think about could I put this to an ASCII file or even pipe
this to another process.
An article on this is at www.oriolecorp.com under PDQOUT.
2. Server concepts explain the function of the rollbacks 3. Not a good idea 4. Try what has been suggested earlier. What I do is provide a tool for mydevelopers which allows them to calculate how many rows before a commit. In a nutshell:
Cheers
Peter Lomax
Kevin Gray wrote:
> Hi there
>
> I am writing some SQL scripts with Unix shell as well to do some very large
> updates (could be moving 3.5million records in insert/select combination).
>
> Can anyone provide answers to the following.
>
> 1. Is it possible to have a single insert/select run without requiring a
> rollback segment to use?
> 2. Is it possible to use multiple rollback segments for a single
> insert/select transaction?
> 3. Is it possible to use multiple files in a single rollback segment so
> that the rollback segment can be split over several files and, possibly,
> disks?
>
> Also, does anyone know exactly the best way to work out how much space to
> allow for rollback segments.
>
> Any help would be very much appreciated.
>
> Thanks.
>
> Regards
>
> Kevin Gray
>
> Kevin Gray
> Senior Consultant
> The Customer Engagement Company, UK
> Email: kevin.gray_at_h2engage.co.uk
> WWW: http://www.h2engage.co.uk/
Received on Fri Sep 04 1998 - 03:37:17 CDT