RE: Any-one know how to eliminate PLANNED downtime with Oracle RAC?

From: Michael Fontana <>
Date: Tue, 25 Nov 2008 10:35:40 -0600 (CST)
Message-ID: <>

Unfortunately, I know of many applications which will fail DRAMATICALLY within a second of database availability interruption.  

Implementing this suggestion would require these applications be stopped, which is often a 30 minute undertaking, followed by an addition 30 minute restart.  

While it is minimal, this is still downtime.  

I suppose it could work if it were well tested with an application that had some built-in fault tolerance for database availability.  

However, I've rarely seen one, and a "minute or two" can equally turn into a longer period as a database shutdown involving "hundreds of thousands" of

In-flight transactions might well turn into 30 or 40 or some other unworkable number.          

From: [] On Behalf Of Yechiel Adar
Sent: Tuesday, November 25, 2008 10:05 AM Cc:
Subject: Re: Any-one know how to eliminate PLANNED downtime with Oracle RAC?   I was in a stream class today and the lecturer mentioned just this thing. Create a second database and create bi-directional streams between the two.

1) bring the application down for a minute or two.
2) change the application to access the second server.
3) bring down the first database.
4) bring up the application. It will start to put updates in the queues in
the second database.
5) upgrade the first database.
6) bring up the first database and wait for the apply process to catch up.
7) bring down the application for a minute or two.
8) point the application to the first database.
9) start up the application.

Upgrade completed with only a few minutes down time.

Need EE for streams and works best in

Adar Yechiel
Rechovot, Israel

Martin Berger wrote:

Hi Keith,  

I have to second Carels and Michaels meanings. Your desire is highly complex and multi dimensional. So you will not get any straight forward answer.  

In one of my prior lives I had to promote and support Multi Master Replication. If someone uses this wise, he can achieve a zero-downtime environment.

But be warned: You need a tremendous engineering work and still really good skilled operational DBAs with enough time to take care of.  

I have never checked, wether or not streams can provide the same functionality. Maybe it's worth checking.  

just some ideas, might they help,



Martin Berger    

Hi, I'm working with a customer running a critical web site on a 10gR2 RAC backend DB - they support hundreds of thousands of simultaneous connections at the "quietest" time.  

They have expressed a desire for NO downtime during ANY changes to Oracle, particularly the application of Oracle patches and Oracle upgrades (both minor and major), etc.  

Any thoughts? Who's "been there done that"?      

-- Received on Tue Nov 25 2008 - 10:35:40 CST

Original text of this message