Benchmarking: iPerf Benchmark Run on ‘m1’ and ‘m3’ Instances – Flux7

By Flux7 Labs
March 18, 2014


In our previous post, we ran benchmark to measure network performance of ‘i’ instances using iperf tool. In this post we will talk about iperf benchmark run on ‘m1’ and ‘m3’ instances. Unlike ‘i’ instances, ‘m1’ and ‘m3’ instances does not support ‘enhanced networking’ which led us to have both the client and the server running in ‘ec2-classic’. However, we have ensured that the machines under test (client and server ) are in the same availability zone.

We ran the iperf tool 10 times to measure the network bandwidth between two machines of same instance type. Out of the 10 results the top two and the bottom two outliers were discarded . We took the average of the remaining 6 results and considered that as our final number. Although 10 is a small number, we wanted to get some quick results. Also since the standard deviation is quite low, we can safely say the the network performance was fairly consistent when the experiments were run.

The table below gives the bandwidth we saw for different instance types.

The chart below gives the graphical view of the the data in the above table. In our experiment we saw that the network performance of m1.large was lower than that of m1.medium. As per AWS documentation, the network performance of both m1.medium and m1.large is ‘moderate’ so we expected it to be similar. We also see that the network bandwidth of m3.medium is less than that of m1.small even though AWS documentation says that the network performance m3.medium is ‘moderate’ while that of m1.small is ‘low’. Network performance often depends on various scenarios like the network load on the actual hardware on which the VM is running. This might be a possible reason for this anomaly.

Another observation is that, the network bandwidth for m1.xlarge,m3.xlarge and m3.2xlarge had a ceiling and were restricted to around 1.09 Gbps. This is evident from the fact that the standard deviation for these instance types is nearly zero.



Did you find this useful?  

Interested in getting tips, best practices and commentary delivered regularly? Click the button below to sign up for our blog and set your topic and frequency preferences.


Categories: Blog
Share This Article