Latency
Network latency
The time it takes for data to travel from your device to a server and back.
Latency is the round-trip time for a data packet to travel from your device to a server and back, measured in milliseconds (ms). It is the number most people call ping. High latency makes interactive applications — gaming, video calls, VoIP — feel sluggish even on a fast connection.
Components of latency
Total latency is the sum of four distinct delays that accumulate along the path:
- Propagation delay — the time for a signal to physically travel the distance between sender and receiver. In fiber, signals travel at roughly two-thirds the speed of light (~200,000 km/s), adding about 1 ms per 200 km of cable. This is the irreducible floor — no amount of hardware can overcome the physics of distance.
- Transmission delay — the time to push all bits of a packet onto the wire. A 1,500-byte packet on a 10 Mbps link takes 1.2 ms to transmit; on a 1 Gbps link it takes 0.012 ms. This component is negligible on modern broadband.
- Processing delay — time spent by routers, firewalls, and switches inspecting and forwarding each packet. Each hop adds a fraction of a millisecond under normal conditions.
- Queuing delay — time a packet spends waiting in a buffer before being forwarded. Under congestion this is the largest and most variable component, responsible for bufferbloat and the difference between idle and loaded latency.
Latency targets by use case
| Use case | Target latency | Why |
|---|---|---|
| Competitive gaming | < 20 ms | Sub-frame timing; every ms affects hit registration |
| Casual gaming | < 60 ms | Perceptible but tolerable lag |
| VoIP calls | < 150 ms | ITU G.114 standard; above 150 ms conversations feel unnatural |
| Video calls | < 150 ms | Same threshold as VoIP; lip-sync issues appear above this |
| Web browsing | < 100 ms | Each page requires multiple round trips; high latency compounds |
| Video streaming | < 500 ms | Buffered delivery tolerates latency; affects startup time only |
Distance and the speed-of-light floor
The minimum achievable latency between two points is set by physics. Light travels approximately 300,000 km/s in a vacuum, but in fiber optic cable the refractive index reduces this to about 200,000 km/s — meaning 1 ms of propagation delay per 200 km of cable. London to New York is roughly 9,000 km of undersea cable, setting a minimum round-trip time of ~45 ms before any routing, processing, or queuing delay is added. You cannot reduce latency below this floor regardless of ISP or hardware. Choosing geographically close game servers and CDN nodes is the only way to reduce propagation delay.
Wi-Fi latency overhead
Wi-Fi adds 1–5 ms of latency per hop under ideal conditions due to the CSMA/CA access mechanism — devices must contend for the channel before transmitting. Under congestion from neighboring networks or many local devices, this can rise to 10–30 ms. A wired Ethernet connection eliminates this overhead entirely and also reduces jitter, producing more consistent latency.
Idle vs loaded latency
Idle latency — measured when nothing else is using the connection — reflects propagation, processing, and transmission delays. Loaded latency — measured while the connection is saturated — adds queuing delay from bufferbloat. A connection with 8 ms idle latency may show 200 ms loaded latency if the router has large buffers and no SQM. This gap is the most meaningful indicator of real-world performance for interactive applications. Run a loaded latency test (not just a basic ping) to understand how your connection behaves when someone is downloading or streaming at the same time.
Tools to measure latency
ping is the most basic tool — it sends ICMP echo requests and reports round-trip time. mtr (My Traceroute) combines ping and traceroute, showing per-hop latency and packet loss continuously. Loaded latency tests (such as the bufferbloat test at waveform.com or a speed test with latency-under-load reporting) reveal queuing delay that idle ping misses. iperf3 with UDP mode can measure one-way latency in controlled environments.
Frequently Asked Questions
What is the difference between latency and ping?
They are the same measurement. Ping is the tool; latency is the metric it measures. Both describe the round-trip time in milliseconds from your device to a server.
Does faster internet reduce latency?
Not necessarily. Latency is determined by distance and routing, not bandwidth. Switching from cable to fiber at the same location will reduce latency (fiber has less signal degradation), but buying a higher-speed plan on the same technology usually does not.
How do I reduce my latency?
Use a wired Ethernet connection (saves 5–20 ms over Wi-Fi), choose game/call servers geographically close to you, and switch to a fiber ISP if cable or satellite is your current technology.