Whose network is traced?
Globalping runs thousands of probes on other people’s home lines and in data centers. Outbound starts from a probe on the same carrier as you, in your city where possible, and shows how a network like yours reaches the target. Return starts from a probe near the target and traces back to your IP. The probe’s own LAN isn’t counted.
Why isn’t the suggested probe in my city?
Many cities have no probe on your carrier. Carriers differ far more than cities do (each runs its own backbone and border links), so the same carrier elsewhere ranks above another carrier in your city. Tap the probe to choose your own.
What does a city with a question mark mean?
Some backbone addresses aren’t announced in the routing table, so a geo database only knows where the block is registered (China Unicom Global’s, for example, in Hong Kong). Such places get a question mark and a dashed line on the map, and they’re left off when the latency to the hops around them rules them out (light in fibre covers about 100 km per millisecond of round trip).
Where do the requests go?
Straight from your browser, never through a 1ms.dog server: to Globalping (the probes and the trace; it receives the target, and your IP on a return trace, and keeps results readable by their ID), to ipip.net, Cloudflare, ip.sb, ipinfo and GeoJS (your exit IPs and where each hop is), PeeringDB (exchange points), RDAP (who registered an address) and public STUN servers.
Why does a hop show loss when the path has none?
Routers answer trace probes last, and often not at all. When one hop drops answers but every later hop is fine, that router was only busy; only loss that lasts to the end of the path counts.
How is the allowance counted?
Globalping counts by the IP that reaches it: 250 traces per hour, one per trace. Behind a proxy it counts the proxy’s exit, shared by everyone on it.
Route Trace