Latency

Latency is the delay between sending a packet and it arriving at the destination, measured in milliseconds. Round-trip time (RTT) — what the ping command measures — is the time for the packet to make the full round trip.

Latency is often confused with bandwidth. Bandwidth is how much data can flow per second; latency is how long each piece takes to arrive. A high-latency link can have plenty of bandwidth — a satellite internet connection might have 50Mbps but 600ms latency. For downloading large files, that's fine. For a video call, it's unusable.

Sources of latency:

Propagation delay — light travels at roughly 200,000 km/s in fiber. A round trip from New York to London is ~10,000km, so physical speed of light alone contributes ~50ms RTT. There's no engineering around this.

Switching/routing delay — each hop through a router adds a small amount of processing time, typically sub-millisecond on modern equipment.

Queuing delay — packets waiting in switch or router buffers when a link is congested. This is where QoS has the most impact.

Serialization delay — time to put all the bits of a packet onto the wire. Negligible at 1G/10G; relevant on low-speed links.

Acceptable latency varies by application. VoIP becomes noticeably degraded above 150ms one-way. Gaming becomes unplayable above ~100ms. Web browsing is tolerant of 200-300ms. Backup and file transfer are throughput-limited, not latency-limited.