Global Ping
Ping any host from probes on six continents at once and see the latency and packet loss your users really experience, region by region, instead of the single figure your own server would report.
About Global Ping
Latency is a property of a path, not of a host. A server in Frankfurt that answers in 3 ms from your office answers in roughly 240 ms from Tokyo and 250 ms from Sao Paulo, and no amount of testing from one location will reveal that. Global Ping sends real ICMP echo probes from measurement nodes spread across Europe, North America, South America, Asia, Africa and Oceania simultaneously, then reports minimum, average and maximum round-trip time plus packet loss for each vantage point, sorted fastest to slowest.
Read the result as a distribution rather than a number. Latency that rises smoothly with geographic distance is simply the speed of light in fibre: about 5 ms per 1000 km each way, so 150 ms between Europe and Australia is physics, not a fault. What deserves investigation is a value that breaks the pattern - one region far slower than its neighbours usually means traffic is being backhauled the long way round, while packet loss above a percent or two on a specific probe points at congestion or a policing device on that path. Loss on the destination itself, reported consistently from every continent, is a problem at your end.
The measurement is the honest way to decide whether you need a CDN or an additional point of presence. If every non-European probe reports 200 ms or more against a single European origin, users in those regions are paying that penalty on every TCP handshake and every TLS round trip, and static assets served from a nearer edge will do more for perceived speed than any amount of server tuning. Conversely, if latency is already reasonable everywhere your users actually are, a CDN adds cost and a cache-invalidation problem for no gain.
Probes come from the Globalping network, an open measurement platform with several thousand nodes in more than 120 countries; our own EU-hosted API orchestrates the run, normalises the results and plots them on a self-hosted world map. Because the hourly measurement budget is shared by every visitor, identical requests are served from a ten-minute cache: checking the same host twice in a row costs nothing and returns instantly. Nothing about your query is passed to an advertising or analytics service, and we set no tracking cookies.
How to use it
- 1Enter the hostname or IP address you want to measure, for example example.eu.
- 2Select Ping worldwide and wait a few seconds while real ICMP probes are sent from every continent in parallel.
- 3Read the three summary cards first: the fastest vantage point, the slowest, and how many probes reached the host at all.
- 4Scan the table for outliers - a region whose average latency breaks the geographic pattern, or any probe reporting packet loss.
- 5Use the map to see the shape of the result at a glance, and re-run after a few minutes if a single probe looked anomalous.
Common use cases
- -Deciding whether a CDN or an extra point of presence is justified, using real per-region latency instead of guesswork.
- -Confirming a new server or anycast node is reachable from the regions it was built to serve, before you send traffic to it.
- -Investigating complaints that a service is slow for users in one country while local testing shows nothing wrong.
- -Checking packet loss towards a host from several independent networks, so you can rule your own connection out as the cause.
- -Producing an independent measurement to attach to a support ticket with a hosting or transit provider.
Frequently asked questions
- Why does ping latency differ so much between regions?
- Because the packet has to physically travel there and back. Light in fibre covers roughly 200 km per millisecond, so a round trip between Europe and Japan cannot be faster than about 130 ms even on a perfect path, and real routes are never straight lines. Compare each probe against its geographic expectation: a value close to the theoretical minimum is healthy, while one much higher suggests the traffic is taking a detour.
- What is an acceptable amount of packet loss?
- For the destination host, zero. Sustained loss of even one or two percent is enough to hurt TCP throughput noticeably, because every lost segment triggers a retransmission and shrinks the congestion window. Brief loss on a single probe can be a transient event on that path, so repeat the measurement before drawing conclusions; loss reported from every continent at once points at the host or its immediate upstream.
- The host does not answer any probe. Is it down?
- Not necessarily. Many operators filter ICMP echo requests at the edge, so a host can serve HTTPS perfectly while never answering a ping. If every probe reports no reply, check the service directly with a port check or an HTTP request before concluding anything: silence on ICMP is a configuration choice as often as it is an outage.
- Can I choose which countries the probes come from?
- This tool spreads the probes deliberately across all six populated continents, which is what you want for a first look at global reachability. Because the measurement budget is shared between all visitors, we keep the selection fixed and cache results for ten minutes rather than allow unlimited custom runs. For repeated targeted measurements from a specific city, a dedicated monitoring service is the right tool.