No announcement yet.

Real world examples of how to use traffic-loopback?

  • Filter
  • Time
  • Show
Clear All
new posts

  • Real world examples of how to use traffic-loopback?

    I was curious if anyone has used the traffic-loopback feature and could give me a real world scenario of how to properly make use of it? I'm stuck in a nearly two month fight with Cogent over a horribly performing 10gig point to point circuit. I know they packetize their p2p circuits, unlike traditional light repeating waves, so I was kind of wondering if I could volunteer to enable "traffic-loopback source network device phy" on my gear on both ends of this link. Then, if they felt like putting effort into this, they could break the point to point at some arbitrary location between the two states I'm connecting, put their own traffic generation equipment on, and see if they can produce loss from that point to either of my locations separately. Or, is there a way I can use the "traffic-loopback source network device phy" feature on one end to more thoroughly test from the other end, whether that is possible within EOS, or via a system on the same vlan, or a system/equipment I hook to the physical cable on the other side? My current config, to facilitate testing, is about as simple as it can get now, where each end is nothing more than a switchport mode access and an assignment to a vlan that has just one virtual machine on each side. I use iperf3 under linux to do both tcp and udp testing, where udp will show bursty loss, and tcp will show bursty retransmits, in between segments of zero loss. Maybe there's some lower level tool I could use on one end, or within EOS, to better test the circuit with it looped on one side. I can't use iperf in EOS because it's cpu-bound and can't generate enough traffic for a 10gig link, but maybe there's some kind of layer two frame generator in EOS that can saturate a looped back link and show me loss. .
    Last edited by kodexof731; 08-15-2023, 12:37 AM.