The Internet is a connectionless network. The routers, switches and all entities on the network are designed for connectionless traffic. This is probably one of the main reasons for the success of the Internet: The routers can be kept simple, nevertheless the technologie scales very well from lowest to highest line speeds.
However, todays link speeds in the gigabit range make it necessary to find more effective ways to handle IP packets in routers and switches. The amount of data transferred by single applications is is growing, and various tasks could be simplified a lot if it would be possible to find a way to aggregate similar kinds of traffic.
Surprisingly, as we will see in this report, the same techniques that are being developed for traffic measurement and analysis will show to provide possibilities to attack this kind of problems as well. As an example for such a development, we briefly present ``flow switching'' in sections 2.5.2 and 3.1.
Traffic aggregation will also be interesting for the deployment of IP traffic onto ATM networks. Here the aggregation of traffic with different characteristics can be used to reserve the appropriate resources and bandwiths automatically [7,11,22].