Determining how much of the network bandwith is utilized during/after a jmeter test

Question

I want to understand how to determine whether my internet connection is a bottleneck or not during a jmeter stress test.

Here is a bit of background for my problem:

I made a stress test with a 2 hours duration, with 2000 threads ramped up over the whole 2 hours duration hoping to find the upper limit of the server, but I am thinking that maybe I hit some other limit and not the server's, as it's not shutting down, aka not responding with 4xx or 5xx error codes. I can only see in the report that the transactions per second go fairly quickly to 580, but they then will not go above.

When I run the stress test against one server endpoint and the load keeps increasing, I need to be connected to a VPN to have access to that endpoint.

Most of the stress test runs I tried, behave like this:

Although, the number of threads increases, like this:

And I only have Non-HTTP error response codes which are not reflected in the server logs, aka the server reports no 4xx or 5xx error codes:

The server will not go over 580 TPS, but it will not throw errors either:

As per this site, my internet connection (via cable or wifi, I see no major difference) has 1ms ping, 400-420 Mbps download capacity and around 250 Mbps upload capacity - granted, this is calculated with some server selected by them, located relatively near to my location.

How can I know how much of that "pipe" capacity is being used so that I can say with certainty that my internet connection is not a bottleneck? Can the VPN be a bottleneck?

Answer 1

There are multiple options like:

1. Bytes Throughput Over Time chart which is a part of JMeter's HTML Reporting Dashboard

   

2. Bytes Throughput Over Time chart from JMeter Plugins project

   

3. JMeter PerfMon Plugin which provides complete Network IO metrics

   

4. Your operating system should provide built-in applications for network monitoring