Re: 2 recovery questions

No, when you perform a recovery in that manner, the controlfile has to be from the "current" point in time. You can't restore the controlfile from a backup and then try to roll forward past the time
(actually sequencing information) recorded in the controlfile. That's
why it is imperative to maintain multiple copies of the controlfile.

To protect against loss of all controlfiles, you should include an "alter database backup controlfile to 'XXXXXXX' reuse" statement as part of any backup strategy. You then issue the "recover database....." command with the "using backup controlfile" syntax added on to it.

In some situations, this will allow you to roll forward past the sequencing information contained in the backup controlfile. Go to technet.oracle.com and review the backup/recover documentation on controlfile recoveries.

It sounds like you are describing an event where you lose the server and all disks that are associated with it. What some would describe as a "disaster recovery event". I agree with the previous poster. You need to consider Oracle Data Guard (previously called Standby Database). It has become quite a popular configuration to protect against disasters.

In its most basic configuration:

The primary database ships redo logs across the network to a duplicate
(standby) database running on another server. The standby database is
in recover mode and the logs are used to keep the database in sync with the primary database. Note that I use the words "in sync" loosely here. Depending on how a Data Guard environment is configured, the standby database will lag behind the primary by some "N" number of minutes. Oracle has made great advances with Data Guard. Please note that there is usually a tradeoff between primary/standby synchronization and primary database performance.

Chris Foot
Senior Oracle Instructor OAEC Program
www.oracleeducator.com
Contemporary Technologies Received on Sun Jan 18 2004 - 09:34:12 CST