On this page
Break down the observed path
The path includes the player, relay and backend. A ping to the relay alone cannot describe the delay to the game. Record where connections terminate or are forwarded. A relay can improve one access network's path while making another longer; group observations by network and region rather than relying on one overall average.
Compare equivalent conditions
Use the same service, comparable time windows and repeated samples. Track loss, delay variation and connection setup along with RTT. Record game-server load to avoid blaming the relay for application processing time. Arrange any direct-origin comparison in a controlled environment instead of leaving the origin permanently exposed.
Observe changes in service
During relay failover or filtering activation, test new connections separately from existing sessions. Their behavior depends on the transport and retained state. Preserve before-and-after observations with timestamps and source networks. This gives the provider evidence about a particular route instead of an untestable claim of zero added latency.
