Latency vs. bandwidth, explained with a garden hose
Bandwidth is how wide the pipe is; latency is how long the water takes to arrive. For most everyday use the second one matters more than people expect.
Internet plans are sold by bandwidth: 100, 500, 1000 megabits. It's an easy number to compare, so it's natural to assume it describes how fast the connection "feels". Often it doesn't.
The garden hose
Think of a hose. Bandwidth is its diameter: how much water can flow per second once it's flowing. Latency is the time between opening the tap and the first drop reaching the end. A wider hose fills a pool faster. It does not make the first drop arrive any sooner.
Loading a web page isn't like filling a pool. It's more like a long series of small cups: fetch the HTML, discover it needs a stylesheet, fetch that, discover a font, fetch that, and so on. Each cup costs at least one round trip. With 40 sequential requests and a 50 ms round trip, you've spent two seconds waiting regardless of whether the hose is 100 or 1000 megabits wide.
Some rough numbers
- Light in fibre covers roughly 200 km per millisecond, so a server 2,000 km away costs at least 20 ms per round trip before any equipment adds its own delay.
- A good wired home connection adds 1–10 ms to the nearest provider equipment. Wi-Fi typically adds another 2–20 ms, more when the air is busy.
- Above roughly 50–100 Mbit/s per person, most households won't notice extra bandwidth in daily browsing. They will notice an extra 50 ms of latency.
What to optimise first
If your connection already delivers comfortable bandwidth, the cheapest improvements are usually on the latency side: use a cable for the desktop that sits next to the router, fix bufferbloat, and get crowded devices off the 2.4 GHz band. Upgrading the plan is the last step, not the first.