According to Wikipedia, torrent is a computer file that contains metadata about files and folders that needs to be distributed. What metadata, in simpler words it is a file with .torrent extension that contains size of the content, name, structure of the content, network locations of trackers but not the file content itself. For example consider you are downloading a Operating System software online using torrent. A .torrent file is downloaded and it contains the size of the operating system files, folder information like names and etc. But note that it does not contain the operating system.
The Real Thing-
Whenever we download a file from internet it is stored in a centralised location on a server or multiple servers. But due to limited bandwidth of each download, bandwidth caps are enforced and periods where system is not accessible. This creates inefficiency when multiple people want to download same file from single source or location. So what torrent does that it virtually sets up server like environment on each users device called as peer-to-peer connection. Now instead of server to user connection there is a user to user connection which is a very efficient way.
For easy Understanding let’s understand it diagrammatically:
Diagram above shows a simple server client interface. Suppose the bandwidth is 1Gbps and it contains a software file of size 1 Gb that user wants to get. Now three users are trying to access or download the same file from same location. Suppose file distribution is even then each user is going to get a max speed of 300-350 Mbps because each user is going to get 1/3rd of the bandwidth. After this bandwidth cap is reached. So this is the scenario. Consider the same interface with 10,000 users. What about the bandwidth! It’s about 0.1024. Each user is allocated a bandwidth of 0.1024 Mbps.
Now what torrent does is that each server is acting as a virtual server. For clear understanding see the diagram below.
Whatever be the size of file no user is going to have a bandwidth cap that was up to the server.
Each file to be distributed is divided into small information chunks called pieces. Downloading peers achieve rapid download speeds by requesting multiple pieces from different computers simultaneously. Once obtained, these pieces are usually immediately made available for download by others. In this way, the burden on the network is spread among the downloaders, rather than concentrating at a central distribution Server or hub.