-4

I have been seeing these questions quite a bit i.e.

"I have a gigabit link and then I ran an Iperf test are my results normal?"

or

"My 100mpbs interface is only sending 30mbps of traffice, is this normal?"

I do not think that these are appropriate questions for the stack exchange. Furthermore, the ANSWERS are usually given by people that are probably unquallified to give experienced opinions on network normalcy. Also, every time a question is asked about an Iperf test, the initial Iperf test is fudged because the OP used the wrong options and/or doesn't understand what they mean.

I would like to label this entire category of questions as off-topic.

Any thoughts?

Quantify actual vs theoretical network throughput

How to calculate throughput if there is network traffic

How to Calculate Throughput in Gbps

Typical throughput on switched network

| |
  • 2
    I think you're missing the point of this website.. – Michael May Dec 12 '13 at 2:18
9

Regarding iperf questions

Your your argument seems to go along these lines:

Because many iperf questions have fundamental misunderstandings, and answers are provided by unqualified people we should make iperf questions off-topic.

I disagree. Network Engineering is here to help people solve practical network engineering problems. If we need to use tools (such as iperf) that are misunderstood by lots of people, those are exactly the kind of questions we want to answer.

If the question has bad assumptions, answer it and show them where their bad assumptions are.

If you think someone's answer is wrong, downvote it. Try to avoid focusing on qualifications; rather look at the quality of their answer and vote accordingly.

Side note:

I'm glad you asked because many people seem to have abandoned Meta Network Engineering. Network Engineering is owned by the community. The way to take ownership of the site is to ask and debate about these kind of questions.

| |
  • It seems like something that should have a wiki written about and then stickied somewhere on the site. Is there a way to do that in a clean manner with the SE structure? – John Kennedy Dec 11 '13 at 14:43
  • 1
    @JohnKennedy, ask your own question and answer it as a community wiki. Just try to make sure your question conforms to normal site guidelines (i.e. not too broad, not excessively subjective, etc...) – Mike Pennington Dec 11 '13 at 15:09
  • 1
    @John Kennedy On EE.SE we wrote some common questions and answered them using the Answer your own question checkbox. After that, such questions ended up in the frequent section, but I don't know how that was done... – AndrejaKo Dec 11 '13 at 16:56
  • 1
    I am going to take the suggestions above and answer my own question to create a guide on troubleshooting network performance using iperf/jperf. – John Kennedy Dec 11 '13 at 16:59
  • Agreed, if the question sucks, downvote it and wait for someone to provide better input. Throughput is a core concept in network design/engineering. – Ryan Foley Dec 16 '13 at 16:23
4

Throughput is one of the major metrics we use to keep networks operating correctly and iperf is one of the tools used to measure this, which makes it on topic in my mind

Of your examples, two are closed, I think one should be open and you are probably right that the remaining one should be closed. Not because it included iperf, but because there really isn't a good question in the post.

Of course, if you feel strongly about any question (whether it is on or off topic), there is always the option for users to vote to close (or reopen for that matter).

| |
0

Like you already elluded to, now might be a good time to create a well-written wiki for the questions to be referenced back to.

iperf provides a critical metric in network architecture.

| |

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .