Re: 'create tablespace' question

"Matthias Hoys" <idmwarpzone_NOSPAM__at_yahoo.com> a écrit dans le message de news:42582dc9$0$1408$ba620e4c_at_news.skynet.be...
|
| "Sybrand Bakker" <postbus_at_sybrandb.demon.nl> wrote in message
| news:j87g51l1af1dd72kctqtfn7bletbqucmm3_at_4ax.com...
| > On 9 Apr 2005 10:34:44 -0700, "mea0730" <mea0730-google_at_yahoo.com>
| > wrote:
| >
| >>If I create a tablespace with the following syntax and run an import,
| >>all the data goes into the last data file. It's not spread across the
| >>datafiles.
| >
| > Well that is really not surprising, isn't it. Normally a tablespace is
| > just filled up, and a second datafile is only being used after it is
| > added. You seem to expect there is a sort of 'software-striping'
| > algorithm when you create multiple datafiles: there isn't.
| > Apart from that, as you are creating all those datafiles on one
| > filesystem, and you seem to be monitoring how far each datafile is
| > filled up: do you want an administration nightmare, or do you just
| > have plenty of time to administer this database, or do you just want
| > make sure you don't have to ever add space again.
| >
|
| Sybrand,
|
| Are you sure this is correct ? I don't have a system for testing right now,
| but I believe that if you create a tablespace with multiple datafiles, and
| then import an object with multiple extents, Oracle will create the first
| extent in the first datafile, the second extent in the second and so on (in
| a sort of round-robin fashion).
|
|
| Matthias
|
|

I didn't check it with import but i made some tests 2 years ago with "standard" processing (insert, alter table allocate extent) and the answer was: it depends on version.
In 8i, first datafile was filled before an extend was allocated in the next file. In 9i, datafiles were used in round-robin fashion.

Regards
Michel Cadot Received on Sat Apr 09 2005 - 14:57:54 CDT