Re: Striping Tables

From: Gary Walsh <Gary.Walsh_at_hds.co.uk>
Date: 1996/11/26
Message-ID: <57f62o$7er_at_willow.hds.co.uk>#1/1


leslie_at_darth.lfso.loral.com (Leslie Carter) wrote:
>
>I have a very large tablespace with about 40 existing tables all in 1
>datafile. These tables are heavily accessed, and I would like to split
>up the tablespace in 3 datafiles residing on different disks in order to
>increase performance. I plan to export the tables drop the tablespace.
>Recreate it with multiple data files. But how can I manage the
>placement of the tables in the different data files?
>
>I can use the alter database command and bring the datafiles offline to
>make them unavailable as the tables are imported, but I prefer to stripe
>the data, have each table spread across the datafiles.

Try the 'parallel=TRUE' option to sqlldr. You'll need to create 3 'infiles' for each table, then load each 'infile' into one of the data files using 'file=<datafilename>'. I use this with Parallel Server on a HDS SR4300 and it works a treat. Of course, if you want to subsequently insert rows then you've got a slight problem (my tables are effectively read-only).

Gary Received on Tue Nov 26 1996 - 00:00:00 CET

Original text of this message