Online Ping Test from Multiple Countries

Enter a hostname or IP address to ping it from real probes in different countries at the same time. Results show the minimum, average and maximum round-trip time and packet loss for each location, so you can tell a slow server from a slow route.

How the global ping works

The test runs on the Globalping network, thousands of small probes hosted by volunteers and companies around the world. Each probe sends four ICMP echo requests and reports the times. Choose Worldwide for five probes picked from the whole Globalping network (usually spread over several continents, but not one per continent), or one region for four probes close to your visitors.

Reading the results

Latency mostly depends on distance: roughly 10 to 40 ms inside a region, 70 to 120 ms across the Atlantic and 150 to 300 ms between Europe and Asia or Oceania. Packet loss above a few percent on one probe often points to that probe’s network; loss from every probe points to the server or its upstream.

When ping fails but the site works

Many servers and firewalls drop or rate-limit ICMP. CSF, for example, can block ping with ICMP_IN or limit it with ICMP_IN_RATE, which shows up as partial loss such as 2 of 4 replies. Use the TCP ping or the website status check to test the service itself.

Online Ping Test from Multiple Countries at a glance

How to use this tool

1. Enter a hostname such as example.com or an IPv4 or IPv6 address such as 203.0.113.10. If you paste a URL, the scheme and path are removed.
2. Choose where to test from: Worldwide uses five probes, a region (Europe, North America, South America, Asia, Oceania, Africa or Middle East) uses four. Probes are picked by the Globalping network, so Worldwide can return several probes in the same area; choose a region when you need a specific one.
3. Press Ping. Every probe sends four ICMP echo requests and the results usually arrive within 20 seconds.
4. Each press runs a new test; results are not cached. You can run 10 ping or traceroute tests per hour. Private and reserved addresses, and hostnames that resolve to them, are refused because public probes cannot reach them.

A hostname is resolved by each probe, not by srvScripts. Sites using geo-DNS, anycast or a CDN can therefore answer from different addresses in different countries, and the IP column shows which one each probe reached.

How to read the results

Field	What it means
Target	The hostname or IP you entered, as sent to the probes.
Probes	How many locations returned a result.
Average latency	The mean of every probe’s average round-trip time, with the fastest and the slowest probe.
Packet loss	none, or how many locations lost at least one packet.
Probe	City, country code, network and AS number of the probe, for example “Falkenstein, DE · Hetzner Online (AS24940)”. The network matters: a probe in a data centre and one on a home connection in the same city can differ by several milliseconds.
Status colour	Green: all four replies. Amber: some packets lost. Red: no usable result.
min / avg / max	Fastest, average and slowest of the four round trips in milliseconds. A large gap between min and max on one probe points to jitter or congestion on that path.
Replies	Replies received out of packets sent, and the loss percentage, for example “4/4 replies, 0% loss”.
IP	The address the probe actually pinged after resolving the hostname.

A red row that says finished means the probe ran the test but got no reply to any packet; the start of the raw ping output is shown next to it. A red row that says failed or offline means the probe itself could not run the test, which says nothing about your server.

Common problems and how to fix them

Every probe shows 100% loss, but the website loads

The server or a firewall in front of it drops ICMP echo requests. Check the firewall on the server:

# CSF: ICMP_IN = "1" allows ping, ICMP_IN_RATE limits it
grep -E '^ICMP_IN' /etc/csf/csf.conf
# firewalld: "yes" means echo requests are blocked
firewall-cmd --query-icmp-block=echo-request
firewall-cmd --permanent --remove-icmp-block=echo-request && firewall-cmd --reload

Windows Server answers ping only when the built-in inbound rule is enabled:

Enable-NetFirewallRule -DisplayName "File and Printer Sharing (Echo Request - ICMPv4-In)"

On AWS, security groups drop inbound ICMP unless you add a rule for it, and other cloud firewalls work the same way. If you would rather keep ping blocked, test the service with the TCP ping instead.

“example.com does not resolve.”

The hostname has no A or AAAA record in public DNS, or a new record has not reached the resolvers yet. Check it with the DNS lookup and, after a change, the propagation checker.

“That is a private or reserved address.” or “points to a private or reserved network.”

Addresses such as 10.0.0.5, 192.168.1.10 or 100.64.0.1 exist only inside a network, so probes on the internet cannot reach them. Ping them from a machine on the same network, or test the public address the service is published on.

“You have run 10 global tests in the last hour” or “The global test network is busy right now”

The first is the per-visitor limit for ping and traceroute together; the second and “The global test network rate limit was reached” mean the shared probe network is busy. Wait and try again. “The test did not finish in time” means the probes were slow to report; run it again.

Latency looks great, but the site is slow

A hostname behind Cloudflare or another CDN is answered by the nearest edge server, so ping measures the distance to the CDN, not to your origin. The IP column shows the CDN addresses. Ping the origin IP directly if it accepts ICMP, and remember that page speed also depends on PHP, the database and caching, which ping does not touch.

What latency to expect, and why

Light in optical fibre covers roughly 200,000 km per second, which works out to about 1 ms of round-trip time for every 100 km of cable. Frankfurt to New York is about 6,200 km in a straight line, so no ping between them can be faster than about 62 ms, and real cable routes are longer. If a probe a few hundred kilometres away shows 80 ms, the traffic is taking a detour or queueing somewhere; a traceroute from that region shows where.

To compare with your own connection, ping from your computer:

# Linux / macOS
ping -c 4 example.com
ping -6 -c 4 example.com     # force IPv6 (Linux)
# Windows
ping -n 4 example.com
Test-Connection example.com -Count 4

Message from your own ping	Meaning
Request timed out. (Windows)	No reply within the timeout: ICMP blocked, host down or packets lost on the way.
Destination host unreachable.	A router, often your own gateway, has no route to the target or the target does not answer ARP on the local network.
Ping request could not find host example.com. Please check the name and try again.	Windows could not resolve the name.
ping: example.com: Name or service not known	Linux could not resolve the name.

Official documentation: Globalping network, RFC 792 (ICMP).

Related guides: CSF DirectAdmin Install: Tune CSF/LFD Safely (Post-1.689) · VPN MTU Fragmentation: Proven Fixes for Slow VPN Throughput.

Frequently asked questions