How to tune FQ_CODEL for IPv6?
-
I have a NetGate 6100 unit running pfSense+ 25.07.
My LAN network is 10 GbE and the LAN connection to the 6100 is a 10 Gbit/s SFP+ uplink from my main switch. LAN is dual stack with internal traffic being mainly IPv6.
My WAN connection is a 2.5 Gbits FTTH connection using PPPoE over VLAN 911. WAN is dual stack with over 50% of my traffic using IPv6. This includes some of my most important traffic.
I've configured FQ_CODEL as per the recipe in the pfSense documentation, with some small tweaks, to try to optimise latency while minimising the impact on throughput. My results have been great for IPv4 where my bufferbloat test score has gone from A/B to A+. However results are not so good for IPv6.
Specifically:
-
Simply having the firewall rule in place to apply FQ_CODEL to IPv6 traffic reduces IPv6 multi-stream iperf3 throughput by ~10% (I do not see any measurable impact on IPv4).
-
My bufferbloat score for IPv6 is only marginally improved.
I've also run my own tests (as opposed to using various bufferbloat test sites) and I see the same results as above; IPv4 is performing really well, IPv6 much less so.
Is it just that FQ_CODEL isn't very good for IPv6? Or are there additional things that I can do to improve things for IPv6?
-