High Availability Requirements/Concepts

We have two geographically dispersed sites. A primary site for day to day production operations and a secondary site for disaster recovery. We are running an application on Oracle9i Release 2 Enterprise Edition for Windows. At each site the Oracle database is configured in a cluster using Oracle Fail Safe.

We have the following requirements:

R1. The database at the secondary site must be a transactionally consistent copy of the production database located at the primary site.

R2. In the event of unplanned outage at the primary site the database at the secondary site must be switched to the production role.

R3. Production operations must be eventually restored to the primary site.

R4. We must be able to manually turn off the "synchronisation" between the two sites to allow the database at the secondary site to become a test environment. It is acceptable in this circumstance that the primary site does not have a standby site for unplanned outages.

My investigation of many discussions, oracle documents and metalink posts shows there are different ways to achieve the above but I am unsure about the following:

Q1. Can R4 (above) be met if using Oracle Data Guard? That is, can you turn a standby database into a test database that can operate independantly to the production database?

Q2. Following on from Q1, is the standby database rebuilt from a backup of production after you have finished testing?

Q3. From what I have read, Advanced Replication does not appear to have been designed for replicating whole databases. It seems to be aimed more at replicating a subset of objects in a database. I don't think it is suitable for the requirements stated above. Do you agree? Received on Sun Apr 03 2005 - 21:54:52 CDT