How has the community scaled their Orion installations? I am currently monitoring 4000 elements with one web/poller server and a standalone database server. I have set my retention settings to 90 days for detailed stats, 180 days for hourly stats, and 3 years for daily stats. This has resulted in the NetPerfMon database being ~250 GB.
I recently opened a support case about a stability issue I've been experiencing in NPM and support was adamant that my DB was too large and that I need to discard a large portion of my historical statistics. I want to keep statistics for 3 years for trending purposes. I'm perfectly fine with buying more disk/CPU/memory as needed but my DBA indicates the current SQL server is holding up just fine. Does anyone else have a NetPerfMon database of this size? For those of you with far more elements monitored, how have you scaled Orion? Have you greatly reduced your retention settings? Have you setup several Orions and used EOC to tie them together? Any other approaches?
If my database is too large for Orion to handle, is there a way to fix this without changing the retention settings for all stats globally? The custom poller detail table is 49 GB and the various wireless tables are a total of 52 GB. I am less interested in long term retention of this data then something like interface utilization or node availability stats which I really need to trend over long periods of time.
