RE: Indexing snapshots

From: Kuijten, F. (Frank) <f.kuijten_at_teknika.btl.akzonobel.nl>
Date: Tue, 29 Aug 2000 16:48:00 +0200
Message-Id: <10603.115739@fatcity.com>

Abiy,

It's because Oracle does not guarantee the sequence in which the rows from the snapshot log are applied to the snapshot. The following situation could occur with an existing snapshot: - First you delete a record.
- Then you recreate the same record. It is placed somewhere else in the table (that means a different rowid !!). Now, when the snapshot gets refreshed, the newly inserted record could be applied before the delete, resulting in duplicate rows.

In Oracle8, with 'primary key' snapshots, the unique index restriction is lifted (AFAIK).

Greetings,
Frank

---

'NT environments also have support for new character sets, and new defined territories in Europe (Kazakhstan, Uzbekistan, Belgium), ...'

• Oracle 8.1.6 New Features Summary

> -----Original Message-----
> From: ALEMU Abiy [mailto:abiy.alemu_at_criltechnology.com]
> Sent: dinsdag 29 augustus 2000 15:45
> To: Multiple recipients of list ORACLE-L
> Subject: Indexing snapshots
>
>
> Oracle says "Do not use CREATE UNIQUE INDEX on snapshots".
> Can somebody
> tell me why ? I couldn't find any explanation for this. But
> I need to
> create unique indexes, for performance reasons. Is there any
> workarounds
> for this. I'm using Oracle 7.3.4 on Solaris 2.8.
>
> Thank you !
>
> --------------------------------------------------------------
> --------------
> --------
> @biy @lemu
> abiy.alemu_at_criltelecom.com <mailto:abiy.alemu_at_criltelecom.com>
Received on Tue Aug 29 2000 - 09:48:00 CDT