iPerf Client Server Different Bandwidth

According to the Paessler site, bandwidth is defined as the amount of data which is able to be transferred from one point to another in a network in a specific amount of time. Usually, it is expressed as a bitrate and measured in bits per seconds. Bandwidth appoints the transmission capacity of a connection.  When determining the quality and speed of a network or the connection of the internet it is a thing which is important.

Talking about Bandwidth, you may want to know some information about iPerf Client server different bandwidth. To be honest, it is a little bit hard to find information about it. When we try to find information about it, here are the things that we found.

In the Stackflow site, there is a question which is related to the iPerf server and client. The question is about iperf server and client differences. On that question, a user named Ben tells that he has set up a network in Mininet which created two nodes with similar bandwidth and delay. The bandwidth is 10Mbps and the Delay is 10ms. When he is using iperf to test it, he does these steps:

  • Begin an iperf server on Node 1 (10.0.0.2) iperf –s
  • Begin the iperf client on Node 2 (10.0.0.3) iperf –c 10.0.0.2
  • The test completes.

Node 2 which is the client displays a bandwidth of 11.2Mbps and a test time of 10.4 seconds. Node 1 which is the server displays a bandwidth of 9.56Mbps and a test time of 12.2 seconds. Both the client and the server display the similar transfer size which is 13.9 Mb.

His question is, is the time difference because of the delay on each individual packet? The TCP window size is 85.6 Kbyte, so adding a 10ms delay to each packet being sent on the network roughly permits for the difference. But, he would have thought the delay would be 10ms. It is because there is a 10ms network delay on both the sending and getting side. Why is this not the case?

From this question, there are two answers. The first answer is from Howard Shane. He answers that the test start or end should be triggered by a timer. The real start or end time should be involved with several signal notifications or handling overhead so that the real time from both sides are different.

The second answer is from Marcus Sandri. He says that unfortunately mininet displays a set of BUGs unsolved. One of its bugs is related to the throughput higher than the bandwidth which is available. Furthermore, he also states that use of bandwidth delay shows be another problem that should be considered. So, he suggests Ben to repeat his experiments with 0ms delay.

In the Server Fault site, there is also a question about iperf client and server. The title of the question is difference in iperf output between client and server. A user named Eric Morand says that he has been playing with iperf for the last 4 hours and he is not able to understand why the results are different in client and server. He shows the iperf output on client and also the corresponding output on server in that question. After showing them, he says that the client is uploading at 1.38 Mbits/sec but the server is getting at 823kbits/sec. What is the explanation about these confusing results?

Then, a user named EEAA answers that for some reasons, the server thinks the test is taking ~8 seconds longer than the client does. With it, it can drop the overall bandwidth from the perspective of the server to 823 Kbps. Jeff LaCoursiere also tries to answer this question. He says that he believes the issue is the client is communicating with a device which has buffered up the outbound traffic. So, the client believes it is done before the server has actually got the last block of data. He has noticed that this a lot with iperf, the second-by-second report of the server is much more stable than the client’s. He also sees that the first second in the client is always the highest throughput which he assumes is filling the buffer of whatever device along the path is doing so. Then, he shows the information about the client and server testing a link which is bi-directional shaped to 7Mbps.

So, if you want to know more about iPerf client server different bandwidth, you may be able to visit some forums or the sites where we found the questions above. There you are able to find some information about iPerf client server different bandwidth. If there is no information about it there, you can make a question about it so that there will be some users who will help you.

If you want to test the available network bandwidth by using iPerf, we have the steps that you are able to follow below.

  • First, you have to download the iperf utility. You are able to get the copy at iperf.fr.
  • On the server which will be getting data, you have to open an elevated command window and then you have to run this command: iperf.exe –s –w 2m
  • Then, on the server which will be sending data, you have to open an elevated command window and then you have to run this command: iperf –c x.x.x.x –w 2m –t 30s –i 1s. You have to replace the x.x.x.x with the IP address of the server from step 2.
  • Now, you have to review the data returned on the server from step 3.

Quoted from Dell site, Iperf is a network testing tool which is usually used by people and it can create TCP and UDP data streams. Also, it can measure the throughput of a network which is carrying them. This tool permits you to be able to set any kinds of parameters which are able to be used for testing a network. Or, it is able to be used alternatively for optimizing or tuning a network. This tool has a client and server functionality and it has the ability to measure the throughput between the two ends, either unindirectionally or bi-directionally. Iperf is an open-source software and it can run on any kinds of platforms such as Linux, Unix and Windows.

Related Posts