This looks really interesting, but I'm worried about the performance hit my routers will take when polling over 4000 OSPF routes.
I had to NPM disable topology polling on my Cisco 6513's due to CPU spikes to 99% causing loss of WCCP whenever the topology was being polled (every 10 hours).
Polling at long intervals seems to defeat the purpose of using this as "real-time" routing info, if you must log in to the router to verify routes anyway.
I use NCM to run a script at 3AM every morning on all my backbone and edge routers to collecting routing & neighbor info.
It overwrites a separate file for each device and is published in the \inetpub\.... directory.
I've added a hyper-link for each file to the device name text box in Atlas.
This allows one to compare differences in neighbors and routing tables if something changes during the day.
Since it uses CLI and not SNMP to grab the tables, it does not add significant CPU overhead when it runs.