On 01.07.22 04:47, John-Mark Gurney wrote:
[...]
Is that expected that I'm not able to get more throughput within the bhyve
network
stack (guest <-> host)? I was expecting way more then 10Gbit/s...
What does CPU usage look like on the host and in the VM? This sounds
like hitting a CPU bandwidth limit in that you're limited by pps...
1) CPU load on the host is not very high (host server is a dual AMD EPYC 7351
16-Core Processor
system):
3333 root 86 0 4230M 1202M CPU8 8 0:33 49.75%
bhyve{vtnet-5:0 tx}
3333 root 87 0 4230M 1202M CPU0 0 1:03 48.67% bhyve{vcpu 0}
3333 root 85 0 4230M 1202M CPU6 6 0:29 45.52% bhyve{vcpu 5}
3333 root 84 0 4230M 1202M CPU5 5 0:47 38.64% bhyve{vcpu 2}
3390 root 4 0 17M 6472K RUN 3 0:19 38.55% iperf3
3333 root 81 0 4230M 1202M CPU4 4 0:26 32.00% bhyve{mevent}
3333 root 30 0 4230M 1202M CPU10 10 0:25 13.44% bhyve{vcpu 1}
3333 root 23 0 4230M 1202M RUN 9 0:08 5.44% bhyve{vcpu 4}
3333 root 23 0 4230M 1202M CPU7 7 0:17 5.26% bhyve{vcpu 3}
2) CPU load in the VM at around 30%
So there should be room for improvements...
