Re: Snapshot using rowid vs Primary Key

From: Michael Oswald <michael.oswald_at_siemens.at>
Date: Mon, 11 Sep 2000 16:10:16 +0200
Message-ID: <8pip48$ika$1@scesie13.sie.siemens.at>

SANDEEP JAIN <sandeepti_at_home.com> schrieb in im Newsbeitrag: x2Yu5.57407$e11.440779_at_news1.rdc1.nj.home.com...
> What is the difference between
> create snapshot <> refresh fast with ROWID ..
> vs
> create snapshot <> refresh fast with Primary Key ..
>
> I am able to understand Primary Key being used to compare rows but how is
> the rowid used ?
>

For the refresh a snapshot log is created, this simply is a table in which all changes to the master-table are stored. On a snapshot-refresh the snapshot-site looks into the log and updates only this data, which are keyed either by primary key, by rowid or both. The rowid is the physical address of a row in the table. But snapshots created with rowid have some limitations e.g. no aggregation funcitons (SUM, AVG, MIN) or DISTINCT or table-joins, which the primary key-snapshots don't have. The rowid is used for backward-compatibility to 7.xx servers, for 8.xx its better to use primary key snaphots.

Mike Received on Mon Sep 11 2000 - 09:10:16 CDT