[ASTF] elephant-flow using pcap?

[Andreas Bourges]

Hi,

...first of all thanks once more for your 1st-class support! Astonishing - again and again and again... Just great.

I've implemented one test-case which is built on the http_eflow.py for ASTF. Generally this performs as expected (23-24 gbps in loopback), but the DUT has problems, because the HTTP-response is too large to be inspected. The customer is asking me, if it's possible to use a larger pcap (e.g nfsv4/tcp) and use that as a basis for generating one large flow. The idea would be to generate a new session based on the pcacp once the first session is closed.

However, I'm not aware of any knob to generate consecutive streams (like we can in STL, where one stream triggers another one...)?! The only thing would be to adjust cps to a very small value, so that the next flow will be generated approimately when the first flow has ended. But that's very rough.

Thanks for any input,

Andreas

[hanoh haim]

You are correct. There is no ASTF action mechanism that trigger one flow from another flow termination. You could reduce the CPS but you still will have a few concurrent flows as the number of TRex DP threads.

EFlow issue is not necessarily related to the inspection, there are many software blocks that could be affected for example distribution to cores won’t work because only one core could handle one flow.

I would try to remove the inspection to verify this.

Thanks,

Hanoh

--
You received this message because you are subscribed to the Google Groups "TRex Traffic Generator" group.
To unsubscribe from this group and stop receiving emails from it, send an email to trex-tgn+u...@googlegroups.com.
To post to this group, send email to trex...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/trex-tgn/bb8d705b-5a0f-4f1f-9ff0-3b7313749d06%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Hanoh

Sent from my iPhone