Why does a site respond to my browser but not to ping?

Many hosts and networks block or rate-limit ICMP while serving HTTP normally. Cloudflare, AWS, and most corporate firewalls do this. No reply to ping is not evidence that a host is down; it is evidence that ICMP is filtered.

What is a good ping time?

It depends on distance. Under 20 ms to a nearby server is good; 100 ms to another continent is expected and cannot be improved, because it is the speed of light in glass. For gaming and calls, consistency matters more, so aim for jitter under 30 ms rather than for a specific latency figure.

Why is my ping from home worse than the numbers here?

These probes run from datacenters with direct peering. Your path adds the last mile, Wi-Fi, and possibly a congested access network. Use these numbers as a baseline for the server side and compare your own result against them to isolate where the extra time comes from.

Can I ping my own home connection?

You can enter your public address, and the result will usually be filtered or lost, because most home routers drop inbound ICMP by default. That is the safe default and not worth changing to make a test pass.

Does ping measure bandwidth?

No. It measures round-trip delay. A satellite link can have gigabit throughput and 600 ms of latency; a slow DSL line can have 15 ms. For throughput use the speed test instead.

Ping Test

Measure latency and packet loss to any host from three regions, with four probes each and the method named per result.

This tool needs JavaScript. It posts one target to our servers, which ask each regional probe in parallel and wait at most six seconds for each. There is no form-only fallback because a plain form submission cannot show partial results.

Where the probes measure from

HYDHyderabad

Four probes per region, fixed. There is no packet count, payload size or flood option, because those turn a diagnostic into a load generator. Ten runs per day per network, one at a time. Report abuse to abuse@whatsmyip.fyi.

Acceptable use

Use this against hosts you own or have permission to test. Sending probes at a third party through this tool to generate load is prohibited by the terms of service.

Rate limits. Ten runs per day per source address, one concurrent run. Past five runs in a day the response is flagged for a bot challenge, which is not yet presented in the browser.
Four probes per region. Fixed. There is no packet count setting, no payload size setting, and no flood option, because those turn a diagnostic into a load generator.
Blocklist. Network operators can have their ranges excluded from this tool. Write to abuse@whatsmyip.fyi.

Report abuse to abuse@whatsmyip.fyi with the timestamp and target.

Reading the three regions together

The pattern across regions tells you more than any single number.

All three high, all three stable. The destination is far from all of them or is slow to respond. Check what it is.
One region high, two normal. A routing or peering problem on that path. This is the case that a single-vantage test misses entirely, and it is common when a transit provider has a congested link.
All three with loss but low latency. Usually ICMP rate limiting at the destination rather than a network fault. Check whether the TCP probe shows the same loss.

Limits

Latency from a datacenter is not latency from your sofa. These numbers describe the path from three well-connected networks to the destination, which is the server-side half of the problem. For the half that includes your own connection, run the browser-side checks on /reachability-check and compare.

About this tool

Enter a hostname or address and we send four probes from each of three regions, then report the minimum, average, and maximum round-trip time and the packet loss for each. Measuring from three places at once shows whether a slow path sits near you or near the destination. ICMP is used where the network permits it, and a TCP handshake to port 443 substitutes where it does not, labelled per result.

How to read the result

Minimum: The fastest round trip observed, and the closest thing to the true propagation delay of the path. Light in fibre covers roughly 200 km per millisecond, so a 100 ms minimum over 500 km means the path is not direct.
Average: The mean across the four probes, which is useful only next to the minimum and the maximum. An average far above the minimum points at queuing somewhere on the path rather than at a longer route.
Maximum and jitter: The slowest probe, and the spread between fastest and slowest. For calls, video, and games the spread matters more than the absolute number, so 40 ms with 5 ms of spread beats 25 ms with 60 ms of spread.
Packet loss: The share of probes that got no reply. Loss on ICMP alone is weak evidence, because routers deprioritise or rate-limit ICMP under load while forwarding data traffic normally, so loss on the TCP probe is the stronger signal.
Per-region comparison: High latency from all three regions points at the destination or its network, and high latency from one points at a routing or peering problem on that path. A single-vantage test cannot tell the two apart.
ICMP or TCP label: Each result names the method it used, because some networks block ICMP echo. Where ICMP is blocked we time a TCP handshake to port 443, which measures the same path but includes any difference in how the destination treats the two protocols.

Questions people ask

Why does a site respond to my browser but not to ping?: Many hosts and networks block or rate-limit ICMP while serving HTTP normally. Cloudflare, AWS, and most corporate firewalls do this. No reply to ping is not evidence that a host is down; it is evidence that ICMP is filtered.
What is a good ping time?: It depends on distance. Under 20 ms to a nearby server is good; 100 ms to another continent is expected and cannot be improved, because it is the speed of light in glass. For gaming and calls, consistency matters more, so aim for jitter under 30 ms rather than for a specific latency figure.
Why is my ping from home worse than the numbers here?: These probes run from datacenters with direct peering. Your path adds the last mile, Wi-Fi, and possibly a congested access network. Use these numbers as a baseline for the server side and compare your own result against them to isolate where the extra time comes from.
Can I ping my own home connection?: You can enter your public address, and the result will usually be filtered or lost, because most home routers drop inbound ICMP by default. That is the safe default and not worth changing to make a test pass.
Does ping measure bandwidth?: No. It measures round-trip delay. A satellite link can have gigabit throughput and 600 ms of latency; a slow DSL line can have 15 ms. For throughput use the speed test instead.

Last reviewed 2026-09-05.