OEM and Management Regions

I have a large distributed environment which I want to manage using OEM. I have servers at a number of different sites, and a central OEM repository in our corporate data centre.

The various sites with managed nodes do not have network connectivity between them, and they have limited connectivity back to the central OEM repository (enough to update the repository via SQL*Net, but not enough for the Agents to communicate to a central OMS - there is NAT in the way).

Reading the documentation, I get the impression that I should be able to use Management Regions to set up a suitable environment. I'd be pleased if someone could point out the flaws in my thinking (or alternatively, confirm that I'm on the right track!!!)

What I'd like to do is to have a single central OEM repository, and a central OMS to which all client consoles connect. None of the agents at the managed nodes will communicate to this OMS.

At each site, I will have an OMS which uses the central OEM repository, and which communicates with all of the Agents on that site. Each site OMS will be in its own management region.

The idea is that I will have a single OMS to which all consoles can connect, and from which all jobs, events, etc can be managed. Even though the box with the central OMS on cannot see notifications from the agents on the managed nodes.

Is this a reasonable setup? Is this the sort of thing for which Management Regions are the correct approach?

If not, what should we be doing? I'm looking for a solution which does NOT involve clients needing multiple OEM Console sessions open, which DOES allow all nodes to be managed centrally, and which deals with the problem that the Agents cannot communicate back to the central OMS, but only to a local OMS on the site network.

Thanks for any help,
Paul Moore. Received on Wed Feb 06 2002 - 08:29:11 CST