TRex for QoS Policy stress testing

[Luis Lhullier]

Hi, guys!

I´m trying to use TRex for QoS Policy stress testing.

I already configure TRex in two different locations (let´s say HQ and brach), where one side is running ASTF in server-only mode and the other is the client side, also running ASTF (of course).

1) The first one is that when running the stream, it starts with high rates, but after some seconds the drop-rate increases and the received traffic on the server side goes to almost zero.

2) The second is the traffic visibility. Is there a way to see the traffic statistics by type or by stream? Either real-time or saving it to a csv?

For example:

Protocol         Bandwidth     Drop-rate

HTTP 1Mbps 256kbps

HTTPS 2Mbps 128kbps

One idea for this second point is to have two TRex instances on the branch site, where one instance will generate traffic to saturate the link and the second one will be used to test each class at a time. This way the droped traffic will be known. Anyway, I´ll need to export the information about the test.

I see that there is a stateful GUI, but the last version is from 2016 so I think it was not updated. Also, I didn´t see any way to export data from there.

Best Regards!

Thanks!

Luis Lhullier

[hanoh haim]

Hi Luis,

1) You should understand the first point first, try to reduce the rate (CPS) and try to find out why you have so many drops. One point about that that the TCP does not have SACK and advance CC so the simulation of the drop is not accurate. We designed this to find the point of NDR (no drop). 

2) You could use counters per tg_name

see https://trex-tgn.cisco.com/trex/doc/cp_astf_docs/_modules/trex/astf/trex_astf_profile.html#ASTFCapInfo

similar to what we have in STL. two instances is not a good option the Console should support showing a limited number of tg_names 

3) We don't have a GUI for stateful - use Console. The way to export data is to use our automation API to send it to a file or realtime graph 

4) You can change the rate dynamically (the CPS) and you can add/remove profile dynamically so you can have 2 separate profile (one with HTTP and one with HTTPS) and then change dynamically the rate of HTTPS profile 

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 view this discussion on the web visit https://groups.google.com/d/msgid/trex-tgn/957bb688-a83c-492e-85fd-0fd37ea1d880n%40googlegroups.com.

--

Hanoh

Sent from my iPhone

[Luis Lhullier]

Thank you, Hanoh!

I'll check it and let you know!

Best Regards!

[Luis Lhullier]

Hello, Hanoch!

Some updates...

Include tg_names in the py file worked... now I can see different "tabs" on the TUI and there I can group different streams. Great!

Regarding the automation API, how can I use it to export to a csv? I didn´t find this in the docs.

Regarding the drops, I got better results but still above the best performance I think... on this server I needed to use the interfaces as linux_interfaces, since DPDK was not working. Do you think this can be the issue?

Also, I´m using IPs addresses in the astf/sfr.py file from the same subnet than the NIC. This is working this way, but I read in one of your past answers in the Google group the subnet which generates traffic should be behind the TREX, like a router. This recommendation still valid?

Sorry for the question storm : )

Best Regards!

Luis Lhullier

[hanoh haim]

1. You should get familiar with automation. The should use the API to get the counters (every x sec) and save them to csv in your format. 

2. Yes, for best performance you should use DPDK, look into the tcp counters to understand the issue

3. Yes, there is a more complex configuration to bypass this requirement (using EMU)

Thanks

Hanoh

To view this discussion on the web visit https://groups.google.com/d/msgid/trex-tgn/8103c973-2885-4d35-aef1-510aed88b826n%40googlegroups.com.