Ping test

Check a public target from a remote probe, or measure this browser’s HTTP timing to this site.

Measurement mode

Configure options

This starts a remote Globalping measurement. Your target and results are public and may remain available for up to six months. Use a target you are authorized to test.

Hostnames are resolved through Google Public DNS first; one public address is selected for the remote test.

One to three remote probes, five packets per probe. TCP uses port 443. No automatic retries; shared service limits apply.

No remote measurement has started.

Remote measurement

Run the tool above to see your results.

Two different latency measurements

  1. Choose the question

    Use a remote public target or this browser’s HTTP path.

  2. Select the origin

    Globalping probe for remote mode; your browser for HTTP.

  3. Collect a short sample

    Five remote packets, or one HTTP warm-up and five measured requests.

  4. Read scope and failures

    Compare timings only with the same origin, method and destination.

Remote ICMP or TCP measures a Globalping probe’s path. Browser HTTP timing measures requests from this browser to this site.
On this page

Learn more about Ping test

Choose a remote ping test or browser HTTP timing

This measurement offers two measurement origins. Remote ICMP / TCP sends five probes through one Globalping probe toward a public target. Browser HTTP timing measures this browser’s requests to this site. Choose the mode that answers your question: an external reachability observation and your browser’s HTTP responsiveness are different measurements.

Read more

Choose a ping test target and address family

Enter an address without a port, or a hostname without a URL scheme or path. Private, loopback, multicast, documentation, and other unsupported special-use addresses are rejected. This measurement is intended for public destinations you are authorized to check. The example button fills 1.1.1.1 and selects IPv4, but does not submit a measurement. You can inspect all settings before any remote diagnostic begins.

For hostnames, our server resolves the selected A or AAAA family through Google Public DNS, follows aliases, validates public addresses and chooses one literal target. Both the entered name and selected address remain visible. Another resolver or later lookup may select a different destination.

ICMP and TCP ping test options

The default measurement uses ICMP. The TCP option targets port 443, with no arbitrary port selection. Each run requests five packets from one probe. ICMP echo handling and TCP connection handling can differ because a firewall or service applies different policies. Compare like with like: changing protocol, target, region, and time together makes a before-and-after result difficult to interpret.

Read your ping test results

Remote reports retain the actual probe city, country, network, ASN, target, protocol and timestamp, plus the public source response. A requested continent does not guarantee a particular city or ISP. Keep these fields with a shared result.

Read more

Round-trip time in a ping test

Minimum, average, and maximum RTT are measured in milliseconds for received replies. Round-trip time includes the outbound and returning journey, plus any handling involved in generating the response. It does not give a separate one-way delay. This measurement preserves a legitimate zero value if the provider reports one; absent timing information is shown as unavailable rather than silently replaced with zero.

Missing replies in a ping test

The test displays packets sent, replies received, and the percentage of missing replies reported for the sample. For a hypothetical five-packet run with four replies, one missing response represents 20%. That percentage describes a very small denominator. It is not proof that one fifth of all future application traffic will be lost. Reply filtering, rate limiting, and destination policy can affect the observation.

If no replies arrive, RTT fields have no valid sample to summarize. The measurement shows “No reply” instead of a misleading zero-millisecond result. Preserve the protocol and actual probe location when investigating. Try the application's normal connection path and consult relevant service monitoring. A single unanswered diagnostic is not sufficient to declare that an entire website or network is offline.

Running, failed, and incomplete ping test measurements

Remote collection waits up to fifty seconds after creation. Stop ends local collection; Clear removes the display. Neither cancels nor deletes a measurement already submitted to Globalping. Uncertain creation requests are not automatically retried.

Browser HTTP timing: warm-up, median and p95

Browser mode sends one unmeasured warm-up request followed by five sequential measured requests, each bounded to three seconds. Timing starts before the typed API call and ends after the small response is decoded. It includes browser scheduling, HTTP transport, server work and decoding. It is neither ICMP RTT nor a test toward an arbitrary website.

The result chart shows actual successful durations and marks failed requests separately. Median summarizes successful samples. P95 uses the nearest-rank method; with at most five successful samples it equals the largest value. This small series is exploratory evidence, not a stable percentile of long-term traffic.

Stop preserves completed HTTP samples and marks collection stopped. Clear discards them, and changing modes removes the previous mode’s working state. Compare runs on the same connection and endpoint; unrelated site traffic or a connection warm-up can change the result.

What this ping test cannot establish

Public ping test results and service limits

Read more

Remote targets and results are public at Globalping and may remain available for up to six months. Hostname resolution also sends the name to Google Public DNS. Browser HTTP mode contacts only this site’s fixed timing endpoint and accepts no arbitrary target.

Rate limits, missing regional probes and provider failures are operational conditions, not target diagnoses. Wait before retrying a remote measurement. Browser HTTP failures remain individual failed samples and never become zero-millisecond replies.

Ping test sources and measurement method

This measurement uses the Globalping API; its integration guidance describes tools built on the platform. ICMP echo behavior is specified for IPv4 in RFC 792 and IPv6 in RFC 4443. Hostname selection uses the documented Google Public DNS JSON interface. Read our methodology for the site's broader distinction between observed evidence and inferred conclusions.

Ping test questions

What is a good ping test result?

There is no universal threshold independent of distance, origin network, destination, protocol, and application needs. A short reply time from a nearby remote probe does not promise the same experience from your device. Use a measurement to compare controlled observations and understand a specific problem. Application timing and actual user experience remain necessary for judging whether a service meets its users' needs.

Why does my ping test differ from a speed test?

A speed test transfers data to estimate throughput under its own conditions. This measurement sends a small number of diagnostic probes and summarizes their replies. The tests may also use entirely different origins and destinations. A connection can have substantial bandwidth alongside noticeable delay or intermittent problems. Keep the measurements' units and paths visible rather than trying to convert one number into the other.

Can a ping test explain where the path changes?

Use online traceroute to inspect reported hops from a remote probe. It can add route observations, including gaps where routers do not reply. A measurement alone has no hop-by-hop evidence. Even a traceroute cannot directly reveal every physical link or identify the cause of a missing response, so combine the report with knowledge of the affected application and network.

online traceroute

How do I share a ping test accurately?

Use Copy report, review the contents, and include your symptom and expected behavior. The export includes the measurement settings, actual probe, available statistics, collection state, source link, and scope notes. Say explicitly that it was run from a remote probe. If collection stopped early, retain that fact. A carefully scoped report is more useful than an isolated number described as the latency of the whole internet.

Is browser HTTP timing an ICMP ping from my computer?

No. That mode times browser HTTP requests to this site. Remote ICMP and TCP modes instead originate at a named Globalping probe. Their origins and protocols differ, so do not compare the numbers as if they measured one identical path.

Does no ICMP reply prove a web server is down?

No. Diagnostic traffic can be filtered or rate limited while an application still answers. Inspect the probe outcome and use a relevant application check before declaring an outage. Missing replies do not explain their own cause.

Is a remote measurement private to this page?

Globalping measurements use a public-results workflow. Review its disclosure and avoid confidential targets. Clearing the local interface or ending the wait does not retract a remote job or copies of its public result.

Can a hostname select a different address from my browser?

Yes. The tool’s documented resolution path and family selection can choose a public address different from your browser’s usual resolver. Save the tested literal address and probe origin so another operator can understand the actual measurement.