Two Ways to Measure Network Performance

Edit: Some links no longer work, although the methods still should assuming ftp ports are open.

Originally posted December 18, 2012 on AIXchange

Note: The next update for this blog will be Jan. 8.

I was forwarded a newsletter that contained a piece on measuring network bandwidth. I’m sharing it here with the permission of the author, IBMer Doug Herman. Doug says he compiled his information from a Steve Knudson presentation at a recent IBM Champions event.

Although I’ve used both of these methods, I hadn’t previously covered them in the blog.

From the newsletter:

            Recently I had a situation where we were being told that network performance was unacceptable from one site to another across a high speed WAN link. After using the ftp method described below, we were able to show that the network speeds were not working as expected across the WAN. It turned out that there was a routing issue with our VLAN; the network admins had it going over a much slower link than the one everyone thought we were using. Once they made the change, it worked as expected.

Two Methods for Measuring Network Bandwidth (subhead)

First Method – FTP

This test is from AIX 5L Practical Performance Tools & Tuning Guide.

ftp> bin

200 Type set to I.

ftp> put “|dd if=/dev/zero bs=8k count 1000000” /dev/null

200 PORT command successful.

150 Opening data connection for /dev/null

1000000+0 records in.

1000000+0 records out.

226 Transfer complete.

8192000000 bytes sent in 70.43 seconds (1.136e+05 Kbytes/s)

Local: |dd if=/dev/zero bs=8k count=1000000 remote: /dev/null

—————————————————————————————-

Second Method – iperf (download)

Server LPAR – “systemX”

> rpm –ivh iperf-2.0.5-1.aix5.1.ppc.rpm

> iperf –s

————————————————————

Server listening on TCP port 5001

TCP window size: 256 KByte (default)

————————————————————

Client LPAR – “systemY”

> rpm –ivh iperf-2.0.5-1.aix5.1.ppc.rpm

> iperf –c systemX

————————————————————

Client connecting to systemX, TCP port 5001

TCP window size: 64.2 KByte (default)

————————————————————

[ 3] local 10.1.1.100 port 55707 connected with 10.1.1.222 port 5001

[ ID] Interval Transfer Bandwidth

[ 3] 0.0-10.0 sec 384 KBytes 314 Kbits/sec

Client LPAR – using three parallel threads

> iperf –c systemX –P3

Client connecting to dettc005, TCP port 5001

TCP window size: 128 KByte (default)

————————————————————

[ 5] local 10.1.1.222 port 37477 connected with 10.1.1.100 port 5001

[ 3] local 10.1.1.222 port 37475 connected with 10.1.1.100 port 5001

[ 4] local 10.1.1.222 port 37476 connected with 10.1.1.100 port 5001

[ ID] Interval Transfer Bandwidth

[ 4] 0.0-10.0 sec 1.20 MBytes 1.00 Mbits/sec

[ 3] 0.0-10.1 sec 2.25 MBytes 1.86 Mbits/sec

[ 5] 0.0-14.9 sec 256 KBytes 141 Kbits/sec

[SUM] 0.0-14.9 sec 3.70 MBytes 2.08 Mbits/sec

(Note: This was originally published in the Power Systems newsletter. It’s produced quarterly and is available to non-IBMers. This Nigel Griffiths post provides some details and tells you how to subscribe.)

Have you used either method, or do you have another way of measuring network bandwidth? Please share your thoughts in Comments.