I don't know if anybody else has a fair amount of network interfaces that connect to devices that you don't care about (ie end user ports)! But with NPM/NTA you have to poll the interface in order for NTA to show you the graphs of the interfaces. This might not be the ideal setup for 200+ user ports on 1 switch. This is an estimated 80 snmps/150 mil seconds with a 2minute poller setting. The work around that I found was on the ports that I don't want to poll, but what flow data from was to Unmange all of the ports until 2020 and set a calander reminder to change in 2015. With the interfaces in the Unmanaged state NTA still gathers/computes the data for the interfaces, thus giving you visibility into the network with out consuming switch resources via SNMP polls.
Hope this helps somebody else.
Instead of unmanaging the interfaces you might want to consider reducing the interface polling interval for those device interfaces instead. Polling them once every 15,30, or even 60 minutes would be more beneficial than simply unmanaging them.