Queue + mangle performance

Hi all,
I am looking to use Mikrotik to shape my user’s IPv4 and IPv6 traffic.
I want user traffic to be limited at our head end router and not at the CPE - CPE is not practical for us as we often serve multiple customers behind a dumb switch behind a single CPE - the CPE only allows interface based rate limiting.

My thoughts are that the head-end router will have IPv4 and IPv6 mangle rules set up to mark a user’s IPv4 and IPv6 traffic with the same packet-mark - a mark unique to that customer. Then I’d set up a queue for each mark to limit user traffic appropriately.

The question is, does anyone have any input on how running 600+ mangle rules (300+ in IPv4 firwall, 300+ in IPv6 firewall) and 300+ simple queues (for 300+ users) on a CCR1009 might go?

Cheers