Right now, having 4000 devices hit the network at the same time and start taking inventory knocks the database server over for a bit. Having the policy stagger throughout the day, or even just a randomized delay over a customizable amount of time (5 hours, for example), would likely eliminate this issue entirely.