Bandwidth/Throughput testing (NTTTCP)
When testing network throughput performance in Azure, it's best to use a tool that targets the network for testing and minimizes the use of other resources that could impact performance. NTTTCP is recommended.
Copy the tool to two Azure VMs of the same size. One VM functions as SENDER and the other as RECEIVER.
Deploying VMs for testing
For the purposes of this test, the two VMs should be in either the same Cloud Service or the same Availability Set so that we can use their internal IPs and exclude the Load Balancers from the test. It is possible to test with the VIP but this kind of testing is outside the scope of this document.
Make a note of the RECEIVER's IP address. Let's call that IP "a.b.c.r"
Make a note of the number of cores on the VM. Let's call this "#num_cores"
Run the NTTTCP test for 300 seconds (or 5 minutes) on the sender VM and receiver VM.
Tip: When setting up this test for the first time, you might try a shorter test period to get feedback sooner. Once the tool is working as expected, extend the test period to 300 seconds for the most accurate results.
The sender and receiver must specify the same test duration parameter (-t).
To test a single TCP stream for 10 seconds:
Receiver parameters: ntttcp -r -t 10 -P 1
Sender parameters: ntttcp -s10.27.33.7 -t 10 -n 1 -P 1
The preceding sample should only be used to confirm your configuration. Valid examples of testing are covered later in this document.
Testing VMs running WINDOWS:
Get NTTTCP onto the VMs.
Download the latest version: https://gallery.technet.microsoft.com/NTttcp-Version-528-Now-f8b12769
Or search for it if moved: https://www.bing.com/search?q=ntttcp+download< -- should be first hit
Consider putting NTTTCP in separate folder, like c:\tools
Allow NTTTCP through the Windows firewall
On the RECEIVER, create an Allow rule on the Windows Firewall to allow the NTTTCP traffic to arrive. It's easiest to allow the entire NTTTCP program by name rather than to allow specific TCP ports inbound.
Allow ntttcp through the Windows Firewall like this:
netsh advfirewall firewall add rule program=<PATH>\ntttcp.exe name="ntttcp" protocol=any dir=in action=allow enable=yes profile=ANY
For example, if you copied ntttcp.exe to the "c:\tools" folder, this would be the command:
netsh advfirewall firewall add rule program=c:\tools\ntttcp.exe name="ntttcp" protocol=any dir=in action=allow enable=yes profile=ANY
Running NTTTCP tests
Start NTTTCP on the RECEIVER (run from CMD, not from PowerShell):
ntttcp -r –m [2*#num_cores],*,a.b.c.r -t 300
If the VM has four cores and an IP address of 10.0.0.4, it would look like this:
ntttcp -r –m 8,*,10.0.0.4 -t 300
Start NTTTCP on the SENDER (run from CMD, not from PowerShell):
ntttcp -s –m 8,*,10.0.0.4 -t 300
Wait for the results.
Testing VMs running LINUX:
Use nttcp-for-linux. It is available from https://github.com/Microsoft/ntttcp-for-linux
On the Linux VMs (both SENDER and RECEIVER), run these commands to prepare ntttcp-for-linux on your VMs:
CentOS - Install Git:
yum install gcc -y yum install git -y
Ubuntu - Install Git:
apt-get -y install build-essential apt-get -y install git
Make and Install on both:
git clone https://github.com/Microsoft/ntttcp-for-linux cd ntttcp-for-linux/src make && make install
As in the Windows example, we assume the Linux RECEIVER's IP is 10.0.0.4
Start NTTTCP-for-Linux on the RECEIVER:
ntttcp -r -t 300
And on the SENDER, run:
ntttcp -s10.0.0.4 -t 300
Test length defaults to 60 seconds if no time parameter is given
Testing between VMs running Windows and LINUX:
On this scenarios we should enable the no-sync mode so the test can run. This is done by using the -N flag for Linux, and -ns flag for Windows.
From Linux to Windows:
Receiver Sender Receiver Sender
ntttcp -r -m <2 x nr cores>,*,<Windows server IP>
ntttcp -s -m <2 x nr cores>,*,<Windows server IP> -N -t 300
From Windows to Linux:
ntttcp -r -m <2 x nr cores>,*,<Linux server IP>
ntttcp -s -m <2 x nr cores>,*,<Linux server IP> -ns -t 300
Sender Receiver Sender