Some graphs on fd.io looks to be incorrect

[baryluk]

https://docs.fd.io/csit/master/report/trex_performance_tests/packet_throughput_graphs/2n-icx-e810cq.html

D.io CSIT-2302.11

Test Report 2023-03-15 07:45 UTC

Number of graphs there show NDR and PDR tests, but results are identical (not similar, data points look to be identical). Either PDR had zero data loss at max line rate, or more likely somebody manually copied files incorrectly, and they got overwritten by wrong source files.

I do not see this mistake on other performance tests, so I think this is a bug.

[pmikus]

Might be. However platform https://docs.fd.io/csit/master/report was superseded by https://csit.fd.io/trending.

docs.fd.io is not maintained anymore.

[vrpolakatcisco]

Most values are identical, but few of them are different.
Recent improvements made it less likely, but it is still possible for PDR to equal NDR.
It took me a while to find the traffic profile where most results have PDR==NDR, nowadays I see that only on AWS [1].

If you look at logs [2] of every trial result, you see the loss counts are either zero or in thousands (or -1, still no idea why).
When there is no load reliably getting loss ratios between zero and 0.5%, PDR should be equal to NDR.

[1] https://csit.fd.io/report/#eNrtk8sOgjAQRb8GN2YMFJCVC5H_MJWOQHg1bUXx6y2EOBBJXLlz0Udybju3c1NtWoVnjdXBCWMnih0WFcJOjn_c2kVVmgUuA6PwAZ2_c33wGuB3DaGboSc9bDgIKUooZAD74AKAJh_2duiUV8jcEkwGjVDDrew03fpRgqi4GaJD4QXqUBEdHRGWeU9s3SeJuUJOamuXkEE9s7D6IBJfFa9RF0-kE7YPxFPbYELLIqaXMzb1KEpGxe-ykP8svmYh31mEyaZpVT3-jzB5ARGK2ww
[2] https://s3-logs.fd.io/vex-yul-rot-jenkins-1/csit-trex-perf-report-iterative-2402-1n-aws/13/log.html.gz#s1-s1-s1-s1-s1-t1-k2-k3-k14