Measure the round-trip time to a server from probes on six continents, and see where it is slow rather than just whether it is slow for you.
Ping measures how long a packet takes to reach a server and come back. It is the most direct answer to "is this host reachable, and how far away is it in network terms" — but run from your own machine it only ever tells you about your own connection.
This runs the same test from probes on six continents at once, so you can see whether a server is slow everywhere or only slow from somewhere. Enter a domain or an IP address to start.
The figure reported is round-trip time: out to the server and back again, in milliseconds. Under about 50 ms feels instant. Around 100 ms is noticeable on anything interactive. Past roughly 300 ms you are almost certainly crossing an ocean to a server with nothing closer to you.
Distance sets the floor and physics enforces it. Light in fibre covers roughly 200 km per millisecond, and real routes are never straight, so London to Sydney cannot be fast no matter what anyone is paying for hosting. What you are looking for is not a small number but a consistent one.
A connection at 200 ms with no loss will feel better than one at 60 ms losing five percent of packets. Loss forces retransmissions, and TCP interprets it as congestion and slows down, so a small percentage does disproportionate damage to throughput.
Any consistent loss above zero is worth investigating. Occasional loss at a single location is usually a busy link somewhere in between rather than a problem at the destination.
Ping uses ICMP, and a great many servers, firewalls and cloud providers drop ICMP as a matter of policy. A host that never answers a ping can be serving traffic perfectly well.
So treat a silent result as inconclusive rather than as a diagnosis. If the website loads in a browser, it is up, whatever ping says about it.
If one region is 30 ms and another is 350 ms, the server sits in one physical place with no presence closer to the slow region. That is completely normal for a single origin server, and it is exactly the problem a CDN exists to solve.
If a host is supposed to be behind a CDN and you still see a wide spread, that is worth chasing: it usually means DNS is handing the same address to every region instead of a regional one. The DNS propagation checker will show you whether that is happening.
A server can answer pings in 20 ms and still take four seconds to produce a page. Round-trip time measures the network path only — not the database query, the template render, or the twelve scripts the page loads afterwards.
Use ping to answer "is the network between here and there healthy". For how fast the site actually feels, that is a different measurement entirely.
It depends entirely on distance. Within your own country, under 50 ms is good. Intercontinental, 150 to 300 ms is normal and unavoidable. Consistency matters more than the raw figure: a steady 200 ms is better than something bouncing between 40 and 400.
Because the host is almost certainly blocking ICMP, which is a common and deliberate security setting. Ping being blocked says nothing about whether the site is serving traffic.
Because each probe is a different distance from the server and takes a different route. That is the point of measuring from several places at once — one number from one machine hides all of it.
No. Latency and bandwidth are different things. A satellite connection can have huge bandwidth and terrible latency; an old DSL line can have low latency and very little bandwidth. For throughput, run a speed test instead.
Sixteen, spread across North America, Europe, Asia, South America, Africa and Oceania, with four packets from each.