Architectures for advance reservations in the internet

Some applications, mainly grid computing or video conferencing require the transmission of large amounts of data or the reservation of a certain amount of bandwidth on the network. Usually the timing parameters such as start and duration or total amount of data for these transmission is known in advance. However, todays networks do not provide means for enabling advance reservations. In this paper, we present two possible architectures for implementing advance reservation mechanisms on IP networks. The architectures are based on integrated and differentiated services networks. We discuss the advantages and disadvantages of both architectures and how advance reservation mechanisms can be implemented in the respective environment.