The tests simulate three different application profiles: 1. the client speaks first (e.g. sends a request) 2. the server speaks first 3. nobody speaks (i.e. a preconnect)