iperf : Network throughput measurement tool

Iperf is a commonly used network testing tool that can create TCP and UDP data streams and measure the throughput of a network. Iperf allows the user to set various parameters that can be used for testing a network, or alternately for optimizing or tuning a network. Iperf has a client and server functionality, and can measure the throughput between the two ends, either unidirectonally or bi-directionally. You can download it in Sourceforce http://sourceforge.net/projects/iperf .  Iperf is a client server program. In our example, let us start "server" iperf at location1 and let the "location2" be the client.

Server side : To start iperf as "server" service, use the "-s" flag.

[root@location1 ~]# iperf  -s
---------------------------------------
Server listening on TCP port 5001
TCP window size: 85.3 KByte (default)
---------------------------------------

To stop the server, send Ctrl+C.

Client side: For client side, iperf has to be initiated with "-c" option. There are lot of other options, which performs different kind of testing. Few examples are explained below.

1) Measuring throughput using default iperf option. This connects to the iperf server mentioned and finds out the network throughput.

[root@location2 ~]# iperf -c location1
---------------------------------------
Client connecting to location1, TCP port 5001
TCP window size: 16.0 KByte (default)
---------------------------------------
[  3] local 201.87.4.3 port 41505 connected with 131.15.17.19 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.1 sec  27.4 MBytes  22.8 Mbits/sec
[root@location2 ~]#

2) Measuring bidirectional throughput usind -d option.

[root@location2 ~]# iperf  -c location1 -d
---------------------------------------
Server listening on TCP port 5001
TCP window size: 85.3 KByte (default)
---------------------------------------
Client connecting to location1, TCP port 5001
TCP window size: 16.0 KByte (default)
---------------------------------------
[  5] local 201.87.4.3 port 41518 connected with 131.15.17.19 port 5001
[  4] local 201.87.4.3 port 5001 connected with 131.15.17.19 port 59284
[ ID] Interval       Transfer     Bandwidth
[  5]  0.0-10.1 sec  27.2 MBytes  22.7 Mbits/sec
[  4]  0.0-10.0 sec  25.2 MBytes  21.1 Mbits/sec
[root@location2 ~]#

3) Measuring multi-thread throughput. Use -P option to specify the number of parallel threads to be started.

[root@location2 ~]# iperf  -c location1 -P 3
---------------------------------------
Client connecting to location1, TCP port 5001
TCP window size: 16.0 KByte (default)
---------------------------------------
[  5] local 201.87.4.3 port 41523 connected with 131.15.17.19 port 5001
[  4] local 201.87.4.3 port 41522 connected with 131.15.17.19 port 5001
[  3] local 201.87.4.3 port 41521 connected with 131.15.17.19 port 5001
[ ID] Interval       Transfer     Bandwidth
[  5]  0.0-10.1 sec  27.9 MBytes  23.3 Mbits/sec
[  4]  0.0-10.1 sec  27.9 MBytes  23.2 Mbits/sec
[  3]  0.0-10.1 sec  27.8 MBytes  23.1 Mbits/sec
[SUM]  0.0-10.1 sec  83.6 MBytes  69.5 Mbits/sec
[root@location2 ~]#

3) Measuring throughput by disabling Nagle's Algorithm.

Nagle's algorithm, named after John Nagle, is a means of improving the efficiency of TCP/IP networks by reducing the number of packets that need to be sent over the network. [ http://en.wikipedia.org/wiki/Nagle%27s_algorithm ]

[root@location2 ~]# iperf  -c location1 -P 3 -N
---------------------------------------
Client connecting to location1, TCP port 5001
TCP window size: 16.0 KByte (default)
---------------------------------------
[  3] local 201.87.4.3 port 41527 connected with 131.15.17.19 port 5001
[  4] local 201.87.4.3 port 41528 connected with 131.15.17.19 port 5001
[  5] local 201.87.4.3 port 41529 connected with 131.15.17.19 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.1 sec  27.8 MBytes  23.1 Mbits/sec
[  4]  0.0-10.1 sec  27.9 MBytes  23.2 Mbits/sec
[  5]  0.0-10.1 sec  28.0 MBytes  23.2 Mbits/sec
[SUM]  0.0-10.1 sec  83.7 MBytes  69.6 Mbits/sec
[root@location2 ~]#

You can even run the client side on regular intervals using cron daemon and plot graphs (using mrtg or rrd) for the throughput. Hope this is helpful.

Vmware Performance tuning Tips

These posts explain about Vmware performance tuning on Linux servers.

• Tips for improving performance on a Linux host
• Vmware IO performance tips

These links explains the tips to tune the disk io by setting read-aheads and memory based improvements by making changes to kernel parameters like vm.swapiness, vm.overcommit_memory, vm.dirty_background_ratio, vm.dirty_ratio. They also suggests some vmware config tweaks like mainMem.useNamedFile. Worth trying if you have Vmware performance issues. But dont try in production unless you are sure of what you are doing.

EMC CLARiiON AX Demo

Gotitsolutions.org has a great demo or EMC CLARiiON AX. You can see the demo here.