The problem is that it doesn't scale linearly. There's some O(n4) algorithm being used in netstat or some kernel syscalls or something. Once you go over 250k, things get _really_ weird.
Re:Http2 explained
by patrickmcmanus on Sunday February 15, 2015 @09:55AM
(#9052732)
try ss -nt instead.. it uses netlink sockets instead of /proc and generally scales much better
# time sh -c 'netstat -tn | wc -l' 486206
real 0m13.538s user 0m1.698s sys 0m10.380s
It still works with a whole lost of connections. (in fairness, only about 130k were connected)