You would have to disable inband on both ends, and even then I wouldn't expect it to go much above 550 or 600MBps on a RFC2544 test.
You said it youself: "This link performs normally with 0.00% packet loss and full duplex layer 3 IP traffic. Between Adva metro Ethernet equipment we can't get a 500 Mbps test to run cleanly." ... but it kinda goes to shit on something like an RFC2544. On another note, what packet size were you using? On Tue, Jan 16, 2018 at 7:05 PM, Eric Kuhnke <eric.kuh...@gmail.com> wrote: > I ended up disabling inband management on the remote end temporarily, and > got the same results... We are likely replacing the radio set with > something that is actually 1GbE line rate and will pass frames sequentially. > > > > On Tue, Jan 16, 2018 at 4:03 PM, Josh Reynolds <j...@kyneticwifi.com> wrote: >> >> Hi. >> >> I actually ran those tests with Performant/Accedian, which caused UBNT to >> put flow control in. Inband management will still cause issues. Your max >> throughput will drop, but it should be stable. I still have the old test >> results around here somewhere. >> >> Remember, these radios have like virtually 0 buffer. They will either pass >> the framea or start taildropping. >> >> On Jan 15, 2018 1:09 PM, "Eric Kuhnke" <eric.kuh...@gmail.com> wrote: >>> >>> I recall the AF team in the Chicago area running Exfo or JDSU RFC2544 >>> tests on some of their test links, and getting them to pass. >>> >>> Have seen the same results with flow control turned on or off. Right now >>> I can't turn off inband management because we will lose visibility into the >>> far end of the radio link, but will try it. >>> >>> >>> >>> On Mon, Jan 15, 2018 at 10:05 AM, Darin Steffl <darin.ste...@mnwifi.com> >>> wrote: >>>> >>>> Siklu would be a better radio for this link. >>>> >>>> I don't recall anyone ever running that test on Airfiber radios from >>>> what I've read. >>>> >>>> On Jan 15, 2018 11:27 AM, "Eric Kuhnke" <eric.kuh...@gmail.com> wrote: >>>> >>>> Same results for sequence errors either with or without flow control >>>> enabled on the radios, and switch ports that connect to them. >>>> >>>> >>>> >>>> On Mon, Jan 15, 2018 at 9:23 AM, Colin Stanners <cstann...@gmail.com> >>>> wrote: >>>>> >>>>> What are the flow control settings? I remember seeing a graph that flow >>>>> control has a big effect on whether AirFibers can hit near the full speed. >>>>> >>>>> On Mon, Jan 15, 2018 at 11:13 AM, Eric Kuhnke <eric.kuh...@gmail.com> >>>>> wrote: >>>>>> >>>>>> Hey Ubiquiti people who might be watching the list, >>>>>> >>>>>> If you're watching the list, please contact me... I have an AF24 link >>>>>> with "perfect" RF characteristics that won't pass a 500 Mbps full duplex >>>>>> RFC2544 test. It also fails MEF y.1731 tests at the same data rate. >>>>>> >>>>>> 1.70 km, -55 RSL symmetric on both sides, radios report 774 Mbps >>>>>> capacity full duplex. >>>>>> >>>>>> We have already replaced a radio on one end, the data cables, tried >>>>>> bypassing surge protectors, and replaced cables. >>>>>> >>>>>> This link performs normally with 0.00% packet loss and full duplex >>>>>> layer 3 IP traffic. Between Adva metro Ethernet equipment we can't get a >>>>>> 500 >>>>>> Mbps test to run cleanly. >>>>> >>>>> >>>> >>>> >>> >
