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

Home -> Community -> Usenet -> c.d.o.server -> Re: Setup of a complete site failover

Re: Setup of a complete site failover

From: Sean M <smckeownNO_at_BACKSIESearthlink.net>
Date: Tue, 16 Apr 2002 15:58:59 GMT
Message-ID: <3CBC4A48.D8912630@BACKSIESearthlink.net>

Knut Talman wrote:
>
> > I know the concept of standby database. The only drawback, if I am right, is
> > that replication is made on the standby database only when a redo log file
> > is full and closed by the master database, which means that the data is not
> > replicated instantly. In the case that the master server is completely out,
> > we may lose changes made during several minutes (maybe hours?) before the
> > crash. Is that right?
>
> Yes, that's right. Thats why I have a cron job (yes, a CRON job) which issues an
> "alter system switch logfile" every 10 minutes.

If you're talking about pre-9i Standby database, then yes, you are correct (though merely switching logfiles every 10 minutes does nothing to *guarantee* that the standby is no more than 10 minutes behind the primary). However, starting in 9i, Oracle introduced a souped-up version of Standby database called Dataguard which offers other options to avoid data divergence and/or data loss between primary and standby. The basic concept is that you have remotely located online redo logs that the standby uses to keep right up to date with the primary. It may be worth investigating a migration to 9i for this feature (speaking here to the original poster).

Regards,
Sean Received on Tue Apr 16 2002 - 10:58:59 CDT

Original text of this message

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