In this video, Tony Fortunato describes testing the throughput of Ubiquiti EdgeRouterX.
A client asked if I could test and determine if the Ubiquiti EdgeRouter X would meet his needs. He needed a router for a site with an internet connection of 10 Mbps and another with 100 Mbps. The only technical requirement is that it has to support Network Address Translation (NAT).
One common methodology for measuring network performance is to use a traffic generator or packet blaster; a receiving device simply calculates metrics and results. All traffic generators are not the same. I prefer ones that generate real packets that require a real response. I believe this two-way conversation is more realistic rather than the one-way blast.
In this video, I use Netscout OptiView XG’s throughput test with the packet sweep option. Using different packet sizes gives me insight into how the device performs in different networks. The packet size range for most Ethernet networks is 64 to 1518 bytes. All I can tell you is that your network will be somewhere in between the two values.
You would need to figure out the approximate average packet size for your specific application or network. Don’t make a common mistake and assume your average value will be the same everywhere in your network; moreover, it will change.
As you will see in the video, I got some interesting results and determined that this router will suit their needs. As a side note, I configured a Ubiquiti EdgeRouter X and have been using it in my office for over a week with no issues.
No matter what network analysis tool you use, it's critical that you document your methodology and equipment setup as well as your results. Understand that every test will have a different result, which is why I prefer to conduct five tests at a minimum. I discard the highest and lowest results and average the remaining three to get a better sample average. Different tools will report different results since they perform tests differently. Some will use UDP, TCP and other might run directly over IP, not to mention there are different ways to calculate the results. The general rule is to stick with the same methodology and tool for your testing.
Once you learn how your tool of choice performs a test, you might want to see if you can change its behavior. For example, you might want to switch from UDP to TCP in your testing, or analyze more than the single TCP stream your tool sends. Finally, there is nothing wrong with using multiple network analysis tools to run your tests. The OptiView XG uses UDP for its tests.