Your question made me think of a few ideas; perhaps not what you're looking for immediately, but some may be of use for you as you expand or clean up or grow your monitoring solutions:
- If you have a thousand servers, I'd think you have a server management solution already in place. That would be the tool to use to deploy the additional IP address for snmp reporting/polling.
- With a thousand servers, I'd be concerned that a single poller isn't big enough to manage all the elements you should be watching. I'm not aware of the potential number of elements any server might have, but if it's less than fifteen I'd be surprised. And if you're hitting 12,000 - 15,000 elements by only monitoring a small fraction of the items that NPM could be watching, you're missing data that can help you resolve issues quickly. One or more additional pollers might be worth considering.
- If these were switches or routers or ASA firewalls instead of servers, I'd use NCM to push the additional snmp IP address into all of them. I haven't managed servers since 2002, though, and I don't know how NCM might be able to make changes to them. For that matter, you didn't mention you own NCM, so this may be moot.
- If your single-poller-limitation is due to budgetary constraints, you may find useful ideas that can help you obtain the funding you need for more/better polling here: A Stratagem For Obtaining Funding For Your Projects
At the moment, my budget only allows for one poller. It's what we used initially in our terrible spec setup and it seemed to kind of do the job required. A while back I spoke to Orion/solarwinds and they said that for the number of nodes/elements we have it would be perfectly fine to use one poller, maybe just increase the resources on the VM.
I will see how it goes as we are polling incorrectly/not polling all elements. Everything was polled as SNMP. This includes the non polled clusters/the non polled services and the non polled SQL databases. I know I need to change some of it, which also brings me to another issues that I need to resolve that I have just remembered: setting up accounts for all of our domains to change the polling method. I haven't ever really done anything like this before so I want to plan for every eventuality before I bosh it and migrate every node onto a new environment.,
Again, I am going to see how it goes. If we can get more customers, then we can hopefully develop the infrastructure. But again, see how it goes.