Test Run # | Started at |
---|---|
Download generated test domains and corresponding delay configurations for dedicated testing:
We setup a server with as many unique IPv4 and IPv6 addresses as measured delays. Each delay has a corresponding IPv4 and IPv6 address bound to all subdomains of a delay-specific domain name. These domain names are authoritative for a set of delay-specific zones. By using tc netem we delay the traffic for the IPv6 address by the configured amount of ms. This enables us to test how resolver deploy Happy Eyeballs.
We randomize domains in order to prevent caching effects on the browsers side. The measurement without randomization shows us if the resolver does any caching or uses performance measurements to select IP versions. Our current implementation does not support dynamic result updates from the server (other tests collect the results on the browser itself). As the outcome of these measurements is only visible on the server side and not on the client, we currently cannot show its results.