NHRP-based interface state control is a fantastic feature that you can use for faster convergence of very large DMVPN networks (as explained in the DMVPN Designs webinar, you can also use it to solve some interesting backup scenarios). We tested it in a network with over 1000 spokes (using ASR1K as the hub router) using very short registration timeouts, and the CPU utilization of the NHRP process rarely exceeded a few percents.
However, the engineer doing the tests forgot to tell me a crucial bit of the puzzle – he had to tweak the NHRP rate limiting (configured with ip nhrp max-send command). This omission was gracefully pointed out by George Mihalachioaie who stumbled across the same problem during his DMVPN deployment.
Summary: Having short registration timeouts in large NHRP networks is not a problem per-se (and NHRP-based interface state control may significantly simplify your design), but you have to know what you’re doing, and change the defaults. The usage guidelines provided with the ip nhrp max-send command should give you more than enough information, and don’t forget that NHRP replies (not just requests) also count as NHRP packets.