STH Help Requested - Next-gen Networking

Patrick

Administrator
Staff member
Dec 21, 2010
12,397
5,575
113
Hi all,

We did an experiment about a year ago setting up T-Rex to do switch/ router reviews. It was good, but scaling to 10G and beyond was getting more expensive.

I am now thinking about just buying a used Spirent.

Seems like that is a pretty big domain area. Do we have anyone on STH familiar with the licensing/ models of these things? I do not want to get something that ends up being useless. Still, if we could get a decently priced 1G/10G testing solution today, that would actually be a big win.

If anyone can help navigate, it would be much appreciated.

Regards,
Patrick
 

Patrick

Administrator
Staff member
Dec 21, 2010
12,397
5,575
113
OK. So I bought a 2017 era C100 for 10GbE testing. We shall see how this goes.

Pretty cool looking actually. Xeon E5-2687W's!
 

eduncan911

The New James Dean
Jul 27, 2015
651
510
93
eduncan911.com
I've spent time trying to research the rabbit hole of Spirent, and still havent found info on what T-Rex is or this C100 beast with 2687Ws (which is what triggered my comment, as I was searching for a couple for sale).

I'm curious with what you trying to accomplish though. You mention switch reviews... 10G... The only thing I can guess is t-tex is some type of data logger for network testing?

Some search results for Spirent brought up testing automation. If that's the case, I can help with any switch configuration and testing as well as bare metal machines at scale (Ansible, Chef, SaltStack, Packer, Terraform, Jenkins, etc).
 
  • Like
Reactions: Patrick

Patrick

Administrator
Staff member
Dec 21, 2010
12,397
5,575
113
T-Rex was basically what Cisco built because they wanted an open-source alternative to things like the Spirent.

I think I bricked the C100 :-(
 
  • Wow
Reactions: eduncan911

eduncan911

The New James Dean
Jul 27, 2015
651
510
93
eduncan911.com
T-Rex was basically what Cisco built because they wanted an open-source alternative to things like the Spirent.
Ah, Cisco T-Rex. Found it and read up on it using Intel's DPDK... And I've been using iperf this whole time!!

I found a Linux installation example, that includes a way to automate with python.


--

You said the key issue with T-Rex was scaling beyond 10G due to cost.

I take it you want at least 100 Gbps throughput, which is scaling up 10x the hardware?

I ask because I found this 100G test with only two machines, each with a single old six-core E5 V2 CPU to hit 100G (tests were on Red Hat, not with Windows DPDK drivers).


Where was your cost bottleneck? I'd like to try to help you tackle the cost, to keep it open source.