Hi Gentlemen,
For reasons that I don't want to go into we are going to be building a temporary database solution that will have 2 Solaris boxes available.

Using Solaris 8 and Oracle 8.1.7:

The Sysadmins are pushing for a floating IP address for the boxes and the same instance name on each server. They would like (in the case of a crash) the IP address to point to the other db server (which has been recovered to the point in time [roughly] that the first db went down) with the same name etc and ta da...

I was thinking of hotbackups in archivelog mode and frequent transfers to the second box.

I assumed so far that it can be done relatively easily but if anyone has a better solution (bear in mind that this is just a temporary solution) I would love to hear it.

Thanks in advance for your help.

Rgs,
Breen.

I forgot to ask - is the solution I was thinking of possible?

Sysadmins: Can't live with them, can't live without them.

There's a couple of things to keep in mind. Your DNS must be extremely reliable to use hostnames in your tnsnames.ora file.

Also, you may want to look at the Network Admin guide for SQL*Net/Net8. You can specify multiple addresses that get tried in succession until they succeed.

What you're describing is basically a poor mans implementation of a standby database. Oracle 8i has some tools to implement a standby database (although 9i is better) and there are significant instructions on Metalink on how to do it. I followed Note 180031.1 and set it up with a 9i database. I think you can do similar for 8i.

Thanking you Marist89,
The reason I'm not going for a standby database solution is that they specifically want both databases to be called the same thing (mainly because our application server is a fool and has difficulty at the best of times).
This limits what I can do alot as far as I can see (I was going to go with MultiMaster Replication), what does a guy do?

Standby database can have the same name as the source database.

Both have to be open at the same time?

Sorry Marist89,
Didn't read your last comment as a question till now.

They don't both have to be open at the same time.

Personally, I think multi-master replication would be overkill. I would setup a simple standby database and write some scripts to automate the switchover. Your standby will be current as of the last log switch which sounds more than "current enough". In 9i, you can configure your standby database for maximum protection which ensures every write to the redo log is written to the standby in real-time.

Thanks for all your help Marist89, I'm going to go with your advice.

Breen.