Re: Row ID

From: Thomas Kyte <tkyte_at_us.oracle.com>
Date: 1999/09/07
Message-ID: <5gXVN9iZvOvnuiQ9s9O08Ym7S5gC_at_4ax.com>#1/1


A copy of this was sent to "Bob Fazio" <bob_fazio_at_hotmail.com.no.spam> (if that email address didn't require changing) On Tue, 07 Sep 1999 10:54:32 GMT, you wrote:

>On somewhat of the same note. Is it possible to split up a query by ROWID.
>I know that parallel query does something along these lines. For example:
>
>select * from mytab where rowid between min(ROWID) and max(ROWID)/4;
>select * from mytab where rowid between min(ROWID)/4 and max(ROWID)/2;
>select * from mytab where rowid between min(ROWID)/2 and max(ROWID)*3/4;
>select * from mytab where rowid between min(ROWID)*3/4 and max(ROWID);
>
>I know there is a problem with not taking into consideration that
>max(rowid)/4 could be less than the min(rowid),
>I was just trying to make my point.

Yes, PQO uses rowid ranges to dynamically partition large tables for parallel operations. It doesn't really 'divide' the rowid but rather uses the extent maps for a table (info from the data dictionary) to create N dynamic paritions each of approximately the same size. It then ships out N queries that each select all of the data in these non-overlapping but totally inclusive rowid ranges....

>
>Thomas Kyte <tkyte_at_us.oracle.com> wrote in message
>news:j1bRN4HaixU2jT6pYC0soaK3FxeW_at_4ax.com...
>> On Sat, 4 Sep 1999 12:41:26 -0400, you wrote:
>>
>> >ROWID identifies the place where Oracle has stored the row. Anything
 that
>> >causes rows
>> >to be relocated makes an old ROWID Obsolete.
>> >
>>
>> No, there is only 1 case where a relocated row will change a rowid. It is
 only
>> in Oracle8i release 8.1. It is when you update a partition key and have
>> explicitly enabled row movement. When you enable row movement in a
 partitioned
>> table, an update to the partition key is processed much like a
 DELETE/INSERT
>> rather then an update.
>>
[snip]

-- 
See http://govt.us.oracle.com/~tkyte/ for my columns 'Digging-in to Oracle8i'...
Current article is "Part I of V, Autonomous Transactions" updated June 21'st
 
Thomas Kyte                   tkyte_at_us.oracle.com
Oracle Service Industries     Reston, VA   USA

Opinions are mine and do not necessarily reflect those of Oracle Corporation
Received on Tue Sep 07 1999 - 00:00:00 CEST

Original text of this message