It is possible to transfer data in a carousel or rolling process via the broadcasting channel. This means that the data is constantly retransmitted at a certain frequency (e.g. 10 times per minute). As soon as the user switches on his device or requests specific data, this is stored in the end device with the subsequent transfer cycle and can be used by applications.
The best-known carousel processes are teletext for analogue television and the DSM-CC data carousel for digital television.
No return channel is required for the carousel process. The data can be extremely current. However it is only suitable for smaller quantities of data and is unsuitable for video clips.
With push processes, current contents are transmitted at regular intervals (e.g. once daily) and are stored in bulk memory in the receiver unit so that they are available to the user on request.
Push processes are possible via all transmission routes. They are also suitable for VoD. Calling up the data is convenient, with the effect being one of high-performance and high quality. However, the contents can only current to a limited degree. Furthermore, bulk storage is required.
With pull processes, the contents are transferred from a server to the respective receiver device on individual demand. Variants which are familiar from the internet include downloads, progressive downloads and streaming.
Possible networks include the internet itself as well as independent networks such as cable networks with a return channel.
Pull processes are the most flexible. They are particularly well suited for connection with extensive electronic libraries, the individual contents of which are only requested by a very small number of users. The disadvantage is the sheer scale of data transferral, e.g. it is difficult to guarantee high quality for VoD.