Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> Re: What not to do in a standby environment ?

Re: What not to do in a standby environment ?

From: Carel-Jan Engel <cjpengel.dbalert_at_xs4all.nl>
Date: Tue, 07 Dec 2004 17:33:29 +0100
Message-Id: <1102437207.5109.12.camel@dbalert199.dbalert.nl>


Mind that shipping every changed block to the standby comes with quite a penalty on bandwith required. You change blocks for online redo's (twice if mirrored), archived redo's (probably also twice), controlfile blocks

        (tripled?), and all the data blocks. All that can easily increase the amount of data transferred by a factor of 10-30 especially when you wnat to do it synchronously! Apart from that I'm told that EMC transfers tracks in stead of blocks. Haven't been able yet to verify that myself. Saving bandwith by not shipping all the redundant copies you configured on the primary violates one of my base-rules setting up a HA environment: keep everything as symmetric as possible. Otherwise you will land in an unknown environment when you activate the standby. That is just asking for more disasters to happen.

Apart from that: Oracle standby/Data Guard gives you the ability to set a delay in log-applying. That helps you to recover from humanoid or logical errors.

Just my $ 0.02 (which is about 0.015 eurocent)

Best regards,

Carel-Jan Engel

===
If you think education is expensive, try ignorance. (Derek Bok) ===

On Tue, 2004-12-07 at 13:36, Scott Dial wrote:

> A few things I might add:
>
> Don't take your database out of archivelog mode (obviously). You'll have to start planning out your upgrades when you move into higher releases a lot more carefully.
>
> We have automated the creation of datafiles, so when one is created on the primary servers, the log eventually gets sent to the standby server. When the db apply process runs and figures out a file is missing, it automatically stops and adds the file, then restarts. Just make sure your unix admins are onboard, because when they add space to local servers they'll need to make sure they're doing the same at your DR site.
>
> If you use EMC, they have some tools (SRDF) that can be used to completely avoid the use of Oracle's DR (Disaster Recovery) products in favor of shipping every changed block to the DR site.
>
> http://www.emc.com/products/networking/srdf.jsp
>
> smd
>
> +++ Prem Khanna J [07/12/04 10:40 -0000]:
> > Hi Friends,
> > I would like to know the list of "NOT TO DO" on
> > Primary database in Standby database environment.
> > And how to rectify, in case , it's done.
> >
> > i have a partial list like ...
> > NOLOGGING , CLEAR ARCHIVE , RECREATE CONTROL FILE,
> > RESETLOGS , ARCHIVE LOG STOP
> >
> > Please do let me know if you have come across
> > anything that's it left out from the docs in case
> > of standby database DONT'S.
> >
> > Note:120855.1 has got some , but it seems to miss
> > things like NOLOGGING operations . If someone has
> > one such list , can you share it with me ?
> >
> > The environment is oracle 8.1.7.4/Solrais 9.
> >
> > Any pointers would be of immense help.TIA.
> >
> > Regards,
> > Prem.
> > --
> > http://www.freelists.org/webpage/oracle-l
>
>
> - smd
> ------------------------------
> http://triquetra.homelinux.org
> AIM Screen Name: orclrox
> ICQ: 320591667
> ------------------------------
> --
> http://www.freelists.org/webpage/oracle-l

--
http://www.freelists.org/webpage/oracle-l
Received on Tue Dec 07 2004 - 10:57:44 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US