← Y2KDASH // field manual · ping explained
> field manual · the ping guide

What's a Good Ping? Ping, Latency, and Lag Explained

Ping is the number a speed test shows after "download" and "upload." Most people ignore it. Of the three numbers, ping is the one that actually determines whether your connection feels fast. A 50 Mbps connection with a 15 ms ping runs circles around a 1 Gbps connection with a 200 ms ping for every real-time activity — gaming, video calls, voice chat, remote desktop. This page explains what the number means, what counts as good across use cases, and what to do if yours is bad.

PUBLISHED 2026-04-22 · ~7 MIN READ · FIELD NOTES FROM Y2KDASH

What ping actually measures

Ping is the round-trip time between your device and a test server, measured in milliseconds (ms). The tool sends a small ICMP "hello" packet; the server echoes it back; the tool measures the elapsed time. A ping of 30 ms means the packet took 15 ms to travel to the server and 15 ms to come back (approximately — paths aren't always symmetric).

The term was coined in 1983 by Mike Muuss, borrowed from sonar. Four decades later it's still the most useful single measurement of a network connection's responsiveness. In modern speed-test language "ping" and unloaded latency are the same thing: the floor of how fast packets can travel when nothing else is happening on the connection.

Typical ranges by connection type

connection typetypical ping (same country)assessment
Wired fiber5–15 msexcellent
Wired DOCSIS cable10–25 msvery good
Wired 5G home internet20–40 msgood
Starlink (LEO satellite)25–50 msusable for everything
WiFi 6/6E (close to AP)+2–5 ms over wiredfine
WiFi (far from AP or crowded band)+15–50 ms + heavy jitterproblem
DSL / legacy copper30–80 msmarginal
Geostationary satellite (HughesNet, Viasat)600+ msunusable for real-time
Mobile 4G40–80 msgood for most, marginal for competitive gaming

Add roughly 50–100 ms for any cross-continent connection — San Francisco to Sydney is ~170 ms at the speed of light, no router in the world can beat physics. The cross-continent add-on is why game servers, CDN points-of-presence, and video-call regions matter so much.

Thresholds by use case

Competitive gaming (FPS, fighting, racing): under 30 ms is excellent; 30–60 ms is playable with minor hit-registration issues; 60–80 ms feels noticeably slower than opponents with better ping; over 80 ms is a handicap in any ranked context. Consistency matters more than raw number: a stable 50 ms beats a 20 ms average that spikes to 150 ms.

Casual gaming (MMO, strategy, turn-based): under 150 ms is fine. Spikes matter less because the games don't react to input in real time. MMO raids break around 200 ms when healer reactions need to be precise.

Video calls (Zoom, Google Meet, Teams): under 100 ms feels natural. 100–200 ms introduces conversational talk-over. Above 300 ms the full-duplex nature of speech breaks down because people can't read each other's cues fast enough. All three major platforms target 150 ms as their ceiling for acceptable experience.

Voice chat (Discord, Mumble, game voice): thresholds similar to video but tighter. Under 80 ms is the goal. Voice is more jitter-sensitive than latency-sensitive; see the jitter glossary entry.

Remote desktop / cloud gaming: under 40 ms for cloud gaming to feel native. 40–80 ms is playable but noticeable. GeForce Now, Xbox Cloud Gaming, and Moonlight all target similar thresholds. Remote desktop (RDP, VNC, Parsec) is more forgiving because the eye tolerates input lag better for mouse-driven work than for fast-twitch reactions.

Browsing: ping above 200 ms starts to feel slow even for reading articles because every link click involves at least one DNS lookup and one TCP handshake (both ping-bound). Modern HTTP/3 reduces this somewhat by consolidating round trips.

What affects ping — in order of leverage

1. Physical distance to the server. The speed of light is the floor. A San Francisco to New York round-trip is ~70 ms on a perfect fiber path. No amount of router tuning beats that. When a game's lobby lets you pick a region, pick the closest one. When a video call defaults to a far region, switch explicitly.

2. Bufferbloat under load. The single most common reason a connection has "bad ping" is that ping is measured while someone in the house is downloading or uploading. A connection with bufferbloat can add 200+ ms to ping the moment any flow saturates the link. See bufferbloat for the mechanism and How to Fix Bufferbloat for router-level fixes.

3. WiFi quality. WiFi adds 2–5 ms and much more jitter than wired Ethernet, and the penalty gets worse the farther you are from the access point or the more devices share the band. For any latency-sensitive activity (gaming, video calls) a wired Ethernet cable is the single biggest improvement. For WiFi specifically, 5 GHz and 6 GHz bands are cleaner than 2.4 GHz.

4. ISP routing. Sometimes your ISP routes traffic through poorly-peered paths. This is mostly outside your control but visible via traceroute: an unexpectedly long first hop or an intermediate hop with much higher latency than neighbors often indicates a routing problem worth mentioning in a support ticket.

5. Hardware bottlenecks. An underpowered router running CAKE or deep-packet-inspection features can introduce processing delay. At gigabit speeds with heavy firewall rules, a 2015-era low-end router can add 5–10 ms of CPU latency. This is rare and only matters for pings already in the sub-30 ms range.

What does not help ping

How to test your ping properly

Single-shot ping numbers from a speed test are fine for a first look. For anything more serious, measure over time: ping samples vary minute by minute. Y2KDASH samples continuously so the distribution becomes visible — not just the average, but the spikes, the time-of-day pattern, the correlation with whatever else is happening on the link. A 10-minute sample beats a 5-second one for understanding how your connection actually behaves.

Separately, ping from the command line still works and is the most accurate way to measure latency to a specific destination: ping 1.1.1.1 gives you round-trip time to Cloudflare's resolver, which is a reasonable neutral benchmark. For path analysis, mtr (or WinMTR on Windows) shows the ping to every hop along the route, which reveals exactly where latency is added.

FAQ

What's a good ping for gaming?

Under 30 ms to a same-region game server is excellent for competitive shooters and fighting games. 30–60 ms is playable for most games. Above 80 ms, rubber-banding and hit-registration problems become noticeable. The most important number is consistency: a stable 50 ms beats a 20 ms average that spikes to 150 ms.

What's a good ping for video calls?

Under 100 ms total round-trip latency feels natural. 100–200 ms introduces noticeable talk-over. Above 300 ms, full-duplex speech breaks down. Zoom, Google Meet, and Microsoft Teams all target 150 ms as their ceiling.

Is ping the same as latency?

Practically yes, technically slightly different. Ping is the name of the 1983 ICMP echo-request tool; latency is the broader term for time-to-traverse. When a speed test reports "ping" it means round-trip time to the test server — a specific measurement of latency. In common usage the terms are interchangeable.

How do I lower my ping?

Three levers. First, physical distance to the server — pick game servers or meeting regions closer to you. Second, eliminate bufferbloat — if ping is fine idle but spikes under load, your router's queue management needs tuning. Third, switch from WiFi to Ethernet when possible.

Why does my ping spike when someone else uses the internet?

That is bufferbloat. When the connection is saturated, packet buffers fill and every packet waits behind the saturating flow. Fix: queue management (CAKE or FQ-Codel) on your router. A healthy connection adds under 30 ms under load; one with bufferbloat can add 200–500 ms. See How to Fix Bufferbloat.

> measure your actual ping distribution
not just the average

Y2KDASH samples ping every 60 seconds while you use the connection. The chart shows the distribution — stability, spikes, patterns — that a single speed-test number hides.

> LAUNCH Y2KDASH →

Related reading: The Speed Test Lie · How to Fix Bufferbloat · Networking Glossary