We manage close on half a million IP addresses using text files and some spreadsheets; We're migrating slowly to a database, and SWO IPAM.
We have over fourteen thousand subnets, and thousands of domains. [the UW is significantly larger than most people realize]; Our scale is such that anything that relies on polling is going to be wildly inefficient, and we have a strong aversion to anything that requires root, telnet, or other similarly insecure technologies. About 18 months ago I did an extensive evaluation of IPAM software and found their features very immature, especially in helping us migrate into them... if we had started with them they might have been functional, but not from where were coming from. Following from that review we realized that our data was poor for integrating into commercial products -- difficulties in loading it was because of things like out-of-zone records, invalid domain names, and other record-keeping data we had embedded in comments in the files. We're taking on IPv6 in a large way, and dealing with IPv6 addresses is a challenge.
Step 1 was to cleanup the data and to that end we are slowly moving our text files into a database. This database will reflect 'how things should be'.
Step 2 is to use that to populate SWO and then use the integration with UDT to determine 'how things are'.