Performance on CCR2216-1G-12XS-2XQ
Hi All I've got my hands on a couple of CCR2216-1G-12XS-2XQ fitted with XQ+85MP01D 100G transceivers. Just doing some initial testing using the built in BTest Server & SpeedTest tools as a bit of initial benchmarking. So far I haven't been able to get up to 100G. The instantaneous throughput I see in the interface list is round 50G for TCP and 75G for UDP. There is absolutely no config on the devices (queues, firewall rules, etc). Interfaces are showing auto-negotiation at 100G full duplex. My guess is that this tool isn't up to the job of maxing out the link but iPerf would probably get it up there. Thoughts? [admin@lab2] > tool speed-test address=192.168.0.1 connection-count=20 user=admin password=admin ;;; results can be limited by cpu, note that traffic generation/termination performance might not be representative of forwarding performance status: done time-remaining: 0s ping-min-avg-max: 42us / 49us / 453us jitter-min-avg-max: 0s / 5us / 405us loss: 0% (0/200) tcp-download: 46.4Gbps local-cpu-load:44% tcp-upload: 37.6Gbps local-cpu-load:31% remote-cpu-load:36% udp-download: 37.7Gbps local-cpu-load:15% remote-cpu-load:30% udp-upload: 45.4Gbps local-cpu-load:36% remote-cpu-load:1 [admin@lab2] > tool bandwidth-test direction=receive address=192.168.0.1 duration=20 user=admin password=admin connection-count=20 protocol=udp ;;; results can be limited by cpu, note that traffic generation/termination performance might not be representative of forwarding performance status: done testing duration: 20s rx-current: 79.2Gbps rx-10-second-average: 71.6Gbps rx-total-average: 56.8Gbps lost-packets: 1110 random-data: no direction: receive rx-size: 9000 connection-count: 20 local-cpu-load: 25% remote-cpu-load: 98%
Hi, Mikrotik say you should not use BTest on the devices actually managing the link See notes here https://help.mikrotik.com/docs/display/ROS/Bandwidth+Test Either way, I think if you are testing 100G iperf is a better tool. Andrew -----Original Message----- From: Public <public-bounces@talk.mikrotik.com.au> On Behalf Of Shane Clay via Public Sent: Friday, February 16, 2024 8:49 AM To: public@talk.mikrotik.com.au Cc: Shane Clay <shane@caznet.com.au> Subject: [MT-AU Public] Performance on CCR2216-1G-12XS-2XQ Hi All I've got my hands on a couple of CCR2216-1G-12XS-2XQ fitted with XQ+85MP01D 100G transceivers. Just doing some initial testing using the built in BTest Server & SpeedTest tools as a bit of initial benchmarking. So far I haven't been able to get up to 100G. The instantaneous throughput I see in the interface list is round 50G for TCP and 75G for UDP. There is absolutely no config on the devices (queues, firewall rules, etc). Interfaces are showing auto-negotiation at 100G full duplex. My guess is that this tool isn't up to the job of maxing out the link but iPerf would probably get it up there. Thoughts? [admin@lab2] > tool speed-test address=192.168.0.1 connection-count=20 user=admin password=admin ;;; results can be limited by cpu, note that traffic generation/termination performance might not be representative of forwarding performance status: done time-remaining: 0s ping-min-avg-max: 42us / 49us / 453us jitter-min-avg-max: 0s / 5us / 405us loss: 0% (0/200) tcp-download: 46.4Gbps local-cpu-load:44% tcp-upload: 37.6Gbps local-cpu-load:31% remote-cpu-load:36% udp-download: 37.7Gbps local-cpu-load:15% remote-cpu-load:30% udp-upload: 45.4Gbps local-cpu-load:36% remote-cpu-load:1 [admin@lab2] > tool bandwidth-test direction=receive address=192.168.0.1 duration=20 user=admin password=admin connection-count=20 protocol=udp ;;; results can be limited by cpu, note that traffic generation/termination performance might not be representative of forwarding performance status: done testing duration: 20s rx-current: 79.2Gbps rx-10-second-average: 71.6Gbps rx-total-average: 56.8Gbps lost-packets: 1110 random-data: no direction: receive rx-size: 9000 connection-count: 20 local-cpu-load: 25% remote-cpu-load: 98% _______________________________________________ Public mailing list Public@talk.mikrotik.com.au http://talk.mikrotik.com.au/mailman/listinfo/public_talk.mikrotik.com.au
Mikrotik say you should not use BTest on the devices actually managing the link See notes here https://help.mikrotik.com/docs/display/ROS/Bandwidth+Test
Either way, I think if you are testing 100G iperf is a better tool.
Using flent[0] as a wrapper for netperf. And being able to graph the test results could be useful, too. Flent is used by e.g. the bufferbloat project to do testing. [0]: https://github.com/tohojo/flent
participants (3)
-
Andrew Oakeley -
netravnen+mikrotiktalklist＠gmail.com -
Shane Clay