SMO Repository Setup


We are in the midst of creating several new Linux database servers for Oracle 11.2 with NetApp FAS2020 storage. NetApp has a plethora of tools for managing the storage of an Oracle environment, and they are all new to us. We have some unresolved questions about where to put the SMO Repository(ies).

I have not recognized the answers from the SnapManager for Oracle Best Practices guide.

I am an Oracle DBA, not a NetApp administrator, and our NetApp admin also has little NetApp experience or training. With that in mind, I hope my post makes sense.

Our target environment:

Two production database servers, each with two application databases.

One QA/development database server, with multiple QA/dev databases cloned from the production databases.

SMO Repository placement:

Since a SMO Repository cannot be used to manage the database that contains it, I assume you need at least two SMO Repositories if all databases are to be managed by SMO Repositories.

I am assuming it will be reasonable to place one SMO Repository within one of the application databases on production server #1 (it would manage all databases except those on production server #1), and another SMO Repository within one of the application databases on production server #2 (it would manage all databases on production server #1).

If the preferred alternative is instead to place the primary SMO Repository in its own separate database, then:

1) what server?

2) what about a second SMO Repository to manage the primary SMO Repository database?

Does a SMO Repository generate much activity in its database?  If the repository is placed within an existing production database, or in a separate database on a production server, is its activity likely to negatively impact performance of the production applications? I am assuming it will not.

What recommendations about our proposed architecture?

Re: SMO Repository Setup

Hi there,

if you have a high-availability environment (RAC Cluster) i´d go for 2 RAC clusters, each holding the repository of the other cluster.

If you have single server instances, i´d go for a dedicated repository server (even a oracleXE on a VM). There is not much load or space involved, each backup takes 100kb data max in the repository.