Message-ID: <F001.005D9B64.20031212163426@fatcity.com>
Date: Fri, 12 Dec 2003 16:34:26 -0800
To: Multiple recipients of list ORACLE-L <ORACLE-L@fatcity.com>
Sender: ml-errors@fatcity.com
Reply-To: ORACLE-L@fatcity.com
Errors-To: ML-ERRORS@fatcity.com
From: Carel-Jan Engel <cjpengel.dbalert@xs4all.nl>
Subject: Re: Oracle Data Guard
Organization: Fat City Network Services, San Diego, California
Precedence: bulk
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"; format=flowed
Content-Transfer-Encoding: 7bit

Comments in-line

At 00:49 11-12-03 -0800, you wrote:

>But the last time I looked at it, you had
>to enable supplemental logging at the
>database level if you wanted to use logical
>standby.  Two side effects -

Yes, indeed, supplemental logging must be switched on.


>1) As you said, you need a unique/primary key,
>and database supplemental logging copies
>such a key into the redo for every change to a row:
>but if there is not uk/pk, then the whole row is copied.

This will, of course, give you significant more redo when rows are long, 
and UK/{K not there.
So, try to find some unique identification, and, if possible, consider 
creating a meaningless PK.
Might be cheaper than having complete rows copied into redo.


>2) The copy into redo is engineered by copying into
>the UNDO first - and since changes to GTTs are
>recorded into the UNDO, this means you get an
>extra volume of UNDO, hence REDO on all changes
>to 'supposed to be low-cost' GTTs.

Sorry Jonathan, might be a stupid question: what are GTT's?


>The other feature of logical standby is that Oracle
>scrapes the redo log to generate LCRs (logical change
>records) which are then checked against your 'Streams
>rule-sets' - and then written into the local database
>for propagation to the remote via AQ mechanisms.

Yes, but this is performed at the standby database, not at the primary. 
IMHO this is the major difference between streams and Data Guard / Logical 
Standby: With streams the LCR is genereated locally and queued to the 
destinations, with Data Guard the redo gets copied to the standby and the 
redo scraping/LCR generations is done overthere. So, apart from all your 
findings about generating extra (possibly/probably significant) UNDO and 
REDO, the primary is only hurt by sending the REDO's to the Standby(s).


>The overheads could be quite significant.

Yes!. So test your individual situation before going to production. It's 
all a trade-off. Streams, Advanced Replication, Data Guard Physical Standby 
with daily refreshes, Data Guard Logical Standby with extra overhead, they 
all com with your Oracle Licences. Probably even Shareplex could help you 
out, but will cost you plenty of money. There is no panacee, you have to 
find the solution that fits your needs AND budget. If your budget is too 
low, you have to adapt your needs, or quality will suffer.

Regards, Carel-Jan Engel

-- There will allways be another 10 last bugs --


>Regards
>
>Jonathan Lewis
>http://www.jlcomp.demon.co.uk
>
>   The educated person is not the person
>   who can answer the questions, but the
>   person who can question the answers -- T. Schick Jr
>
>
>One-day tutorials:
>http://www.jlcomp.demon.co.uk/tutorial.html
>
>
>Three-day seminar:
>see http://www.jlcomp.demon.co.uk/seminar.html
>____UK___November
>
>
>The Co-operative Oracle Users' FAQ
>http://www.jlcomp.demon.co.uk/faq/ind_faq.html
>
>
>----- Original Message -----
>To: "Multiple recipients of list ORACLE-L" <ORACLE-L@fatcity.com>
>Sent: Tuesday, December 09, 2003 9:59 PM
>
>
>Hi Vi,
>
>Rows NEED unique identification. So, if there are bunches of raw data with
>no unique identifier whatsoever (remember, rowid is not allowed) LSB can't
>generate a where-clause what row to update or delete on the SB database.
>It's generating SQL based on redolog info, and has to come up with an
>UPDATE <table> SET .... WHERE <unique id> = <unique id>. The unique id may
>be a multi-column key. There is an escape. Enabling supplemental logging
>can add extra info to do the unique identification, when no usable keys are
>available. This will cause some extra logging to be generated, there ain't
>no such thing as a free lunch. For detailed information read chapter 4.1.5
>& 4.1.6 in the Oracle Data Guard Concepts and Administration manual, part
>no. A96653-02.
>
>regards, Carel-Jan
>At 15:54 8-12-03 -0800, you wrote:
>
>
>--
>Please see the official ORACLE-L FAQ: http://www.orafaq.net
>--
>Author: Jonathan Lewis
>   INET: jonathan@jlcomp.demon.co.uk
>
>Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
>San Diego, California        -- Mailing list and web hosting services
>---------------------------------------------------------------------
>To REMOVE yourself from this mailing list, send an E-Mail message
>to: ListGuru@fatcity.com (note EXACT spelling of 'ListGuru') and in
>the message BODY, include a line containing: UNSUB ORACLE-L
>(or the name of mailing list you want to be removed from).  You may
>also send the HELP command for other information (like subscribing).

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.net
-- 
Author: Carel-Jan Engel
  INET: cjpengel.dbalert@xs4all.nl

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru@fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).