Advanced TLS Testing Customisation - crt26/PQC-LEO GitHub Wiki

The TLS performance testing script supports several configuration options that allow the benchmarking process to be tailored to specific environments. These options are beneficial when operating in restricted networks, virtualised environments, or when precise control over test behaviour is required.

Supported customisation features include:

• TCP port configuration
• Control Signal Behaviour
• Disabling Automatic Result Parsing
• Enabling Energy Usage Testing

If the default TCP ports are incompatible with the testing environment, custom ports can be specified at runtime using the following flags. Port values must fall within the range 1024–65535.

When using custom TCP ports, please ensure the same values are provided to both the server and client instances. Otherwise, the testing will fail.

By default, the tool uses a 0.25 second delay when sending control signals between the server and client instances. This is to avoid timing issues during the control signal exchange, which can cause testing to fail.

If the default delay is unsuitable for your environment, you can either set a custom delay or disable it entirely:

Please note that the --control-sleep-time flag cannot be used with the --disable-control-sleep flag.

The performance testing script triggers automatic result parsing upon test completion. This behaviour can be disabled by passing the following flag at runtime:

./pqc_tls_performance_test.sh --disable-result-parsing

Disabling automatic parsing may be appropriate in scenarios such as:

• Collecting raw outputs for batch processing at a later stage

• Running tests in low-resource environments

The TLS performance testing script can be configured to collect energy usage metrics during testing. This is done by passing the --enable-energy-testing flag when launching the script on the client machine. Please refer to the PQC Energy Usage Testing Guide page for further details on performing TLS handshake energy usage testing.

The TLS handshake energy usage testing functionality can be enabled by passing the following flag when launching the testing script on the client machine:

./pqc_tls_performance_test.sh --enable-energy-testing

When energy usage testing is enabled, the client machine will gather a baseline measurement for 60 seconds before beginning the TLS handshake tests. This duration can be configured by passing --energy-baseline-duration=<TIME> alongside --enable-energy-testing when calling the client machine script. The specified duration must be a positive whole number of seconds. The script will report an error if the baseline duration flag is passed without energy usage testing being enabled.

An example of using this flag to gather a baseline measurement for 120 seconds is as follows:

./pqc_tls_performance_test.sh --enable-energy-testing --energy-baseline-duration=120

During client configuration, the script asks whether the TLS handshake performance results should also be written to the client machine. Selecting no collects only the energy usage results on the collection machine. Selecting yes additionally stores and parses the handshake performance results on the client; TLS speed testing and parsing remain disabled. Writing the performance results to disk occurs inside the measured test windows and will therefore be reflected in the collected energy usage metrics.