This document uses the example of Bill Gates purchasing Google to explain the difference between bandwidth (or throughput) and latency.
Bandwidth is the rate at which information can be moved, whereas latency is the amount of time between request and response. To show how this translates into real life, consider the following example: Suppose that Bill Gates wished to buy all of the information stored by Google, and he wanted it by tomorrow. So, spending his fabled $30 billion dollars in one transaction, he says "Give me Google's hard drives." And suppose the founders of Google decided to say "OK". Then theoretically all of those terrabytes of data could be loaded onto a truck and delivered to Bill Gates within a day. Suppose that the actual amount of data was 80,000,000,000,000 bytes, and suppose the number of seconds is 86,000. Then the throughput is about 930 Megabytes per second. 930 Megabytes per second sounds like a fast Internet connection. For example, if you were able to download your files at 930 MB/sec, you would be able to download the Microsoft Windows XP setup CD in less than a second. But since the latency was 86,000 seconds, then it would take you 86,000 seconds before the download would start. In terms of web browsing, a latency of 86,000 seconds implies that you would have to wait a day between each click on your web browser for any web page to come up (even very small ones).
The total download time is therefore equal to latency time + ( amount of data / throughput ). This means that if the latency is poor, then the Internet will seem slow. But, even with 0 latency, if the connection throughput is only 300 bytes / second, then a small 50k (fifty kilobyte) document will take 171 seconds (almost three minutes) to load.
