How latency reveals where a VPN server really is
Why a fast reply proves that a server is near, how this archive turns ping times into distance limits, and the rules that decide whether a location matches its country.

Key points
- Nothing travels faster than light in fiber. So the time a reply takes sets an upper limit on how far away the server can be.
- This archive allows 100 km per millisecond of round-trip time: a reply after 10 ms cannot come from more than 1,000 km away.
- A fast reply can rule a country out. A slow reply rules nothing out, and no single reply can prove where a server is.
The physics in one paragraph
Light in glass fiber travels about 200 km per millisecond. A ping goes to the server and back, so it covers the distance twice. That gives 100 km of distance per millisecond of round-trip time. Real cables do not run in straight lines and routers add delay, so the true distance is usually much shorter than this limit. For internet routes, which run through fiber, the limit is never too short, and that is what makes it useful.
From one probe to a finding
One probe is not enough. Probe locations come from databases and can be wrong, and a single odd reply proves little. The rules of this archive:
- Many probes, many places. Each location is pinged from probes in several countries and networks.
- A margin for imprecise probes. Every probe gets an extra 100 km on top of its limit.
- At least 2 independent probes must rule a country out before the status is “server outside the country”.
- A second day. If the provider does not mark the location as virtual itself, a second day must show the same result.
- Conflicting probes are dropped. Two probes that reach the same server cannot be farther apart than their two limits added. If they are, the probe with the most conflicts is discarded.
What “matches the country” means
A location matches its country when no probe rules the country out and the closest reply places the server within 1,000 km of it. That is strong evidence, not proof: the server could still stand just across a border.
If no probe is close enough, the result is “no conclusion possible”. Across the archive:
- matches the country: 644 (57 %)
- no conclusion possible: 346 (31 %)
- server outside the country: 132 (12 %)
Cities, too
Where a provider publishes the coordinates of a city, the same test runs against a circle of 50 km around it. A server in another city of the same country still matches the country; the city result is shown next to the location.
What is measured is the server address the provider publishes for its apps. The address your traffic exits through can differ and would need an account to check. The method lists every limit.
Next step
Read Reading measurement evidence to understand the table of probe replies behind each status.
The figures in this guide come from the archive and change with every data update. As of October 17, 2026.
Check it with the archive's data
- Virtual VPN locations: how many and which providers
- VPN services and providers: locations, measurements, prices
- Method
Read next: Reading measurement evidence: probes, limits, and open resultsAll guides