Benchmarks

This page collects benchmark results which are used in development. We would like to know whether IPFire got faster or slower with certain changes and therefore we collect this data to test IPFire against it whenever significant changes have found their way into the distribution. This will allow the development team to stay on the correct course and avoid introducing any performance regressions.

Setup

These benchmarks have been conducted with IPFire Appliances from Lightning Wire Labs to which the IPFire development team has easy access and which allow us to reproduce the benchmarks regularly.

The setup is as follows:

• All traffic is being passed through the device that is being tested. That means it is connected to two more appliances which pass traffic between them.
• Traffic is being generated with fireperf
• All traffic is being sent in full duplex mode (-D)
• IPFire has to be set up in a default/minimal configuration with no other services besides those required for the benchmark and all essential operating system services being running
• Ten connections are running simultaneously to take advantage of any load-balancing (-P 10)

The fireperf command line looks as follows: fireperf -c 192.168.1.1 -D -P10

Results

Values in brackets show the change to the previous benchmark of the same category.