Skip to content

Full access for 3 days at 1 ₽ — no auto-renewal.

Try for 1 ₽

Tools · Latency

Connection latency

The round-trip time of a request to the server. Conversation liveliness and game responsiveness depend on it — the megabits in your plan don't help here.

HTTP response time is measured, not ICMP ping — ICMP isn't available from a browser, so the number here is always slightly above console ping. The first request of the series is discarded: it's spent on connection setup.

There is one measurement point for now, so there's nothing to compare. More points will appear as they come online.

Latency is not speed

Speed answers “how much data per second”, latency — “how long one packet travels there and back”. The two are independent: a road of any width doesn't get shorter for being wide.

Files care about speed; conversation and games care about latency. A hundred-megabit channel with three hundred milliseconds of latency downloads beautifully and calls painfully: the other person's reply arrives a third of a second late, and people start talking over each other.

Why the headline number is the minimum

Anything can lengthen a packet's trip: a router queue, noisy Wi-Fi, a phone's power saving. Nothing can shorten it: you can't go below the signal's travel time through the cable.

So the series minimum is the most honest estimate of the clean road, while the average is always spoiled by random delays. The median next to it answers “how typical is that minimum”: close to it — the channel is steady.

Jitter matters more than the average

Jitter is how much the samples dance around each other. A steady ninety milliseconds feels better than jumps from forty to two hundred at the same average.

Voice and video buffer the incoming stream, and the buffer size is dictated by the worst case. A steady channel — small buffer, lively talk; a jumpy one — the buffer grows and replies lag. Jitter on the order of the latency itself almost always means Wi-Fi or congestion, not the server.

What latency is made of

The main contribution is distance: in fibre a signal covers roughly two hundred kilometres per millisecond, and a trip to another continent is never cheap. Then come the number of hops, channel load and the last mile — most often home Wi-Fi.

Hence the order of action: your side first — cable instead of Wi-Fi, closer to the router, no parallel downloads. Then compare the points: a fast nearby one and a slow distant one is geography, not a malfunction.

Questions and answers

Different measurements. Ping sends a service ICMP packet; the browser makes a full HTTP request the server has to accept and process. The extra milliseconds are the method's price, and it's stated in the method note.
Up to 60 ms a conversation is indistinguishable from in-person; up to 120 ms the delay is felt but tolerable. Beyond that it starts to interfere, and past 250 ms people talk over each other. For reading pages and watching videos these thresholds barely matter.
The network is a shared medium with ever-changing load. A few milliseconds between runs is normal; a difference of several times is a reason to look at Wi-Fi and parallel downloads.
Yes, always: the route lengthens by the intermediate server. The size of the addition depends on whether the server lies along the way or off to the side.
No. An honest speed test pushes tens of megabytes through the channel; running that on every page open is not something we do.

The latency add-on depends on the point

In MeerGuard White you choose the connection point — the shorter the detour to it, the smaller the addition.

Open the Telegram botif Telegram opens for you
Sign up on the siteif Telegram is blocked