24

Maybe I am missing something rather basic, but I don't know so I figured I would ask.

In terms of testing, how sensitive various TDMoE are to BER and packet delay, I am looking into how to simulate delay and random packet dropping at the ethernet level. For these tests, I will probably just run iptables as documented in a similar StackOverflow question. However, although that will work here, I am not sure it will work in a more general case (perhaps testing BER tolerance between Windows servers and the like). So this has got me thinking.

Are there any inexpensive network testing devices that allow one to artificially delay or drop packets in order to stress test protocols and evaluate sensitivity to latency and packet loss?

My requirements are to be able to do this on the ethernet level. I am not looking so much for specific devices, brands, etc. so much as "are their things that do this?" and if so, maybe some general ideas of what they are like. I am mostly looking at building a simple bridge running Linux and iptables to do this. Just wondering if this is the right way to go.

2
  • "Are there any inexpensive network testing devices" - Be careful, that is basically a "shopping" question, which isn't allowed. Consider revising into something like "Tips for finding an effective networking loss tester" or similar.
    – jwbensley
    May 19, 2013 at 10:01
  • When you say you want to simulate this at the Ethernet level, how will you check that? Ethernet has no TTL, sequence numbers, handshake, options or other connection tracking features, it is up to the higher layer protocols to track this.
    – jwbensley
    Jun 24, 2014 at 12:32

2 Answers 2

26

There are tools like WANem which allow you to simulate WAN links by artificially causing arbitrary delay and loss rates on a link. WANem works at a relatively high-level; you won't see physical errors on the link but packets will be dropped. It can be deployed on commodity hardware.

I know there are a few other tools which serve similar purposes but I can't recall the names of any at the moment.

3
  • 1
    I have had good luck with WANem as well. I installed it on a lab VMWare ESX server so that a quick vSwitch config change could move it around the various lab segments.
    – vitisimus
    May 9, 2013 at 14:05
  • 1
    you could also do this with a random linux-box and iptables. Example: code.nomad-labs.com/2010/03/11/… May 10, 2013 at 7:57
  • 2
    FYI, WANem is just a pretty GUI wrapper around the Linux netem module of the tc (traffic control, the Linux QoS subsystem admin tool.) Do check out the links in @petrus answer. Dec 12, 2013 at 4:12
5

WANem is great, like Jeremy said. Other interesting tools are netem or dummynet.

See the following examples with netem:

# tc qdisc change dev eth0 root netem loss 0.1%
# tc qdisc change dev eth0 root netem delay 100ms 10ms 25%

You can grab more informations on the netem page or on this excellent SO answer.

Not the answer you're looking for? Browse other questions tagged or ask your own question.