When someone says that their computer is slow, it isn’t actually the speed they are complaining about. They say the computer is slow because it is unresponsive. The PC may be fast enough speed wise to compute all of their needs in a timely fashion, however when they try to go open another program, they find themselves waiting. The same concept applies to network latency.
The speed rating of your internet connection is based on its maximum throughput under ideal conditions. As much as I dislike using pipe analogies when talking about network connections; it really is the most ideal in this situation. If you think about a pipe with water flowing in it with a valve in the middle, the amount of water flowing when the valve is open is the bandwidth. Now, let’s say the valve opens and closes a lot, because it is dispensing very specific amounts of water.
When you click on a link, the valve opens, but the amount of time your message to open the valve takes to get to the valve is called your Latency. The time it takes to get to the valve and the water to start coming to you is called the ping time, or round trip latency.
How this impacts bandwidth is a little devious, as well as potentially crippling. Bandwidth is measured as “data measurement / time measurement” in this case its Mb/s. Latency increases the amount of time it is taking to get the data from the server to you. If you had a 1.5Mb satellite connection, its average round trip latency is somewhere in the neighborhood of 1000ms-1500ms. You add this to the bandwidth equation to get what the connection really feels like in terms of speed.
Taking 1.5Mb or 1536kb and dividing it by 2 seconds instead of 1, (1000ms = 1 s) gives us an average bandwidth of 768Kb/s. Dividing it by 2500ms (initial 1 second measurement + latency) gives us a speed of 614Kb/s. Of course that is also on a single load, if you are using a website that uses multiple loads when you make selections in drop down lists this penalty is applied every time the page loads. To give a real world example, my parent’s satellite internet was 768k and had an average latency of 1800ms. 768/2.8 = 274Kb/s, and it felt like it.
Now sure, if you are downloading large files that don’t change often, the speed picks up. If it was only 274k a download would top out around 35KB/s.(Kb/8 = KB) Theirs topped out around 100KB/s but the actual experience of using it was abysmal. You were lucky if you could even get to a YouTube video, much less watch it. Everything nowadays has been organized into small “web ready” packets for easy streaming. This approach cripples a high latency connection.
This wasn’t meant as a knock on satellite internet. It serves an important niche where the only other choice may be dial up. Just remember though, speed is not always speed, and responsiveness is one of the most underrated metrics in the world.