Round trip
The number that matters first. Measured over thirty seconds and reported as a median, not a best case.
Forty two backbone locations across five regions. The list below shows what each region offers and the kind of drop a well-routed session sees. Your own result will depend on your line and the game server.
| Region | Nodes | Typical drop | Best for |
|---|---|---|---|
| North America | 14 | -38 ms | East and west coast servers, US and Canada |
| Europe | 12 | -44 ms | Frankfurt, Amsterdam, London, Paris hubs |
| Asia | 9 | -51 ms | Singapore, Tokyo, Seoul, Mumbai |
| Oceania | 4 | -29 ms | Sydney and Melbourne, cross Tasman routes |
| South America | 3 | -33 ms | Sao Paulo and Buenos Aires |
A node twenty kilometres away on a congested exchange can lose to one two hundred kilometres away on a clean backbone. That is why selection is measured end to end.
The number that matters first. Measured over thirty seconds and reported as a median, not a best case.
The spread between packets. A route with a flat median beats a faster one that bounces, because the game can predict it.
Any loss on the path disqualifies a node for that session, regardless of how good its latency looks.
Nodes outside a practical radius of you and the game server are dropped before any probing, which keeps the first scan under thirty seconds.
Each candidate is scored on median latency, jitter and loss. Loss is weighted hardest, latency next, jitter last.
The top candidate runs a longer confirmation probe. If it holds, the session moves. If it does not, the next candidate is tried.
The app does the comparison. You just need to be on the connection you play on.