Re: the 20% rule

From: David BL <davidbl_at_iinet.net.au>
Date: Sun, 27 Jul 2008 17:47:27 -0700 (PDT)
Message-ID: <18231f32-e5da-4716-9be6-c46dcb09a708_at_x29g2000prd.googlegroups.com>


On Jul 26, 1:59 pm, Tim X <t..._at_nospam.dev.null> wrote:
> David BL <davi..._at_iinet.net.au> writes:
> > On Jul 24, 10:01 pm, aark..._at_gmail.com wrote:
>
> >> The fastest way to retrieve rows from a table is to access the row
> >> with exact row id. An
> >> index is the second fastest way
>
> > I'm no expert in oracle, but my understanding is that it provides so
> > called heap organised tables or else index organised tables. The
> > above comment would seem to apply to the former not the latter.
>
> Oracle provides a wealth of options with respect to table organisation,
> indexes etc. Rowid is very fast because it identifies the exact row in
> the table to retrieve. The problem is that you rarely know the rowid
> without first having selected the row - its not like using an index
> where you can say "give me all records where the field x = y". However,
> if you do know the rowid of the row you are interested in, it is very
> fast because you access it directly from the table and not by first
> reading the index to get the table location and then reading it from the
> table. At least, thats sort of my understanding in a siplistic way.

AFAIK in an index organised table, the ROWID is actually the primary key which is used as a primary index. Your comment suggests that an index is necessarily a secondary index which isn't always the case. Received on Mon Jul 28 2008 - 02:47:27 CEST

Original text of this message