Effective bandwidth is the actual speed at which data can be transmitted on a connection. This is as opposed to the theoretical maximum that the connection can carry. Originally effective bandwidth was used in reference to dial-up devices, though the term can also be used for permanent connections such as broadband Internet providers.
An example of effective bandwidth being limited is with home broadband services. Each customer connected to a local exchange will be given a maximum possible speed which is available on their connection. However, if every customer connected at this speed at the same time, the amount of bandwidth that passed through the local exchange would be very and extremely expensive to provide. This is highly unlikely to happen, so the connection is not set up for it.
The broadband provider or local phone operator will instead limit the total bandwidth. The limit chosen will be a balance between the highest demand it expects to receive and the desire to keep costs low. This usually means there will be situations, for example at peak times, when the total bandwidth demanded by customers exceeds that which has been made available. This means that customers will suffer slowdowns and their effective bandwidth will be noticeably slower than their maximum bandwidth.
However, there are many other reasons beyond artificial limits why an effective bandwidth may be lower than the stated maximum. The very simplest is that data will be slower to transfer if it has to travel further. This can mean the effective bandwidth varies between users in different locations.
Another issue is the way data is sent in packets. These are small batches containing part of the data being sent, plus information to identify the sender and recipient, and an error checksum, which is the mathematical equivalent of a checklist to make sure the data has arrived safely. At every junction in the data’s journey around the network, the entire packet must arrive before it can be checked and then routed on the next stage of its journey.
The delay between the first part of the packet arriving and the last part arriving slows down the overall speed. This means the effective bandwidth is inevitable slower than the theoretical maximum. There is also the possibility of delay when two or more packets arrive at the same gateway simultaneously, meaning they effectively have to form a queue, which is also known as a buffer.