I do a test following customer env, and the steps are recorded here: https://github.com/wangzheng422/docker_env/blob/dev/redhat/ocp4/4.14/4.14.net.observ.md . The RTT test is ok for several curl request. But when we do load test, we find the traffic number is not the same with pod metrics. It should be 10MB, but the network observability reports only 15KB. The problem is disccused in thread: https://redhat-internal.slack.com/archives/C01DMAD88JF/p1713963069451079

1. create a test app which will call backend service. The test app will have a random delay, then call the backend service.
2. create a backend service out side of the ocp. This backend service can handle concurrent request.
3. create some traffic to the test app with 1000 concurrent request, can call backend with 8k file.

the network observability reports around 15KB, but the traffic is about 10MB

the number will be the same with pod metrics.

siege -q -c 1000 http://wzh-demo-llm-demo.apps.osp-demo.wzhlab.top/sendRequest