Standby Database

The below is a note from technet..
Implement a managed standby configuration, whereby a standby database automatically applies archived redo logs that are automatically shipped to the standby site by a primary database. In this way, changes to a primary database are regularly propagated to a standby database.

How do we automatically apply archived redologs??

Thanks

Not sure. Seems you are talking about the future. Thats Oracle 9i. I don't remember exactly but, its analogous to the concept of intelligent agent wakes up frequently and apply the logs. I meant strictly analogous/similar not at all intelligent agent concept. I don' have something in mind to compare the concept for explaination purpose.

This is Oracle 8i. You modify the initsid.ora and specify the archive destination for the standby database. The detailed steps are included in the Standby Database Concepts and Administration Guide for your version.

Method1:
rely on Oracle's subsystem to move and aply logs to standby database via net8

Method2:
1. move logs to standby server with script (or just do not move, but map disk with primary archived files (if on NT) and point standby's init file log_archive_dest_1 to point to maped disk)
2. aply archived logs with script, execute the script with chron job (at command for NT)
Standby db with time lag can only be implemented with scripts (see sample perl script in Oracle's manuals) For NT I use javascript on NT javascript engine.

For detail refer the standby db manual, and also refer the adminstration manual for setting up your archive directory.

Good luck,
Sam

Thanks to all of you.. Great response.. This feature is there fully supported from 7.3.4 onwards and 8.x with very few restrictions.. I shall catch up with the doc..

I am confused. Yes, what I have mentioned in my Posting is With respect To Switching Stand by when Primary DB is down.
Nothing to do with applying logs. Sorry!

If you are using 8i and configured primary and standby database in Unix environment, then Oracle uses rpc to copy the archived log file from primary DB to standby DB. For which the location where archvied files are copied in standby DB must be specified in init.ora file of the primary DB.

The only problem I had faced was when ever Oracle fails to copy the archived files into stand by db server, a huge number of files are left out in the Primary DB. Hence, I discontinued this automated process of copying.

A cron job will do the same and it will be under your control. Please remember that the cron job should not copy half filled archived files into stand by db. If it is done, then your stand by DB will become useless. While reading the files from the archvied directory, you should omit the latest archived file(s). The number of files being written by Oracle is based on the number archive processes running the primary DB.

I think it works like this:

- Two DB set up identically
- One is open and active, in ARCHIVELOG mode with automatic archiving
- One of the archive log destinations is a location which is set as the archive log destination for the standby DB
- The standby DB is constantly in recovery, thus is not open for transactions but instead is continuously applying the logs stored in the directory it shares with the primary DB
- If the primary DB goes down, merely end the recovery on the standby DB and open it for connections/transactions.

Of course, use of the standby database will result in lost transactions, even committed ones if the primary DB goes down. This is because the online redo logs might not/won't be available to the backup DB. Better to use OPS to avoid lost transactions . . .

Is this right?

[Edited by kmesser on 03-07-2001 at 09:02 PM]

Can anyone shed any more light on this subject?