How we measured
Globalping sent 16 ICMP packets from each of three selected probes in 16 cities to 37.235.48.250. The run started at 09:42:18 UTC and finished at 09:48:20 UTC on 18 September 2026. All 48 requested ping probes returned results.
For each city, the displayed range covers the three per-probe average round-trip times. It is not a minimum-to-maximum packet range or a forecast of customer latency. The expanded rows identify the network and ASN, mean RTT, individual packet extremes and packet loss.
MTR measurements followed the pings using the same probe selections. The raw outputs are provided for route inspection. Missing intermediate-hop responses can reflect ICMP filtering or rate limiting; they are not, on their own, proof of end-to-end packet loss.
What the sample does—and does not—show
The ping target is the published Warsaw Looking Glass endpoint, not an identified customer VPS plan. These inbound IPv4 measurements do not benchmark virtual CPU, disk I/O, application response time or IPv6.
766 of 768 ping packets were received. The Frankfurt OVH probe and Paris Scaleway probe each lost one of 16 packets (6.25% in that individual sample). This short observation cannot establish long-term uptime or sustained packet-loss rates.
Routes differ substantially: the three Vilnius probes averaged 9.149–51.293 ms, and the London probes averaged 29.066–86.324 ms. The full table keeps those differences visible.
The Ookla tests ran on host kvm-lps-pl05. The GSL download phase reported 16.22 ms jitter and a 210.99 ms maximum latency; the 0.38 ms headline is idle latency. Each result describes one server and one moment, not a per-VPS guarantee.
The Ookla tests are dated 18 September 2026. The Globalping run has recorded start and finish times in UTC. Results are a snapshot; test your own route before ordering.