The Traditional Way

When you usually download a file from the internet, the download program connects to a remote server, sends your request and then the remote server sends you the file.

This transfer method, used by the HTTP and FTP protocols, is called client-server model, because the central server keeps the file prepared for others to download it, there is a permanent dialog and connection between the client and the server. The method is very good when downloading small files but you really get to see its negative site when you try to download a large, popular file, requested by lots of people.

For example, if you start downloading a recently released CD image of a Linux operating system you will get a reasonable speed at the start, but, as time goes by and more users request that file from the central server, the speed drops because the central server has to divide its attention and Internet connection with all the users, in rare cases it can even crash under the load. Also, as all people are downloading from the central server, if there is a networking problem or a hardware failure, all 300 people would remain with partial downloads because they all rely on this server to send them the file. This is partially solved with download mirrors but this is often not a viable solution due to the high costs in network connectivity, hosting and equipment.

The central server (in the middle) sends the file to each computer (client-server model)

Most people don't think about how much bandwidth is used by this central server. In most countries, companies do not pay by the minutes the computer is online (as in Dial-up), they are charged for the data that is sent through the line (usually fiber optics). For a large company, 1.4 GB (2 CD Images) times 300 (users downloading the images) may not be much but for a small company or a simple person, this is very hard to swallow.

This page was viewed by 824 people. It belongs to the All about BitTorrent guide, listed in the internet category. Print this guide.