Almost all traffic in todays networks can be classified as being either
stream or elastic traffic. Stream traffic is generated by time-sensitive
applications like VoIP, video etc and has strict bandwidth and delay
requirements. Elastic traffic on the other hand is generated by
applications such as file-transfers where the total download time is
more important than inter-packet or end-to-end delay. The support of
these two traffic types is possible either with a Differentiated
(DiffServ) or an Integrated Services (IntServ) architecture. However,
both DiffServ and IntServ rely on efficient scheduling mechanisms and
control to meet the requirements of these traffic types. In order to
meet the strict delay requirements of stream traffic, the choice of
priority scheduling in inevitable. The danger with priority scheduling
lies in its potential starvation of lower priority traffic classes for
example the elastic traffic class. Therefore, it is essential to
understand and capture t!
he relation between stream and elastic traffic behaviour, priority
scheduling and traffic control. In this paper we focus on developing
simple yet effective approximations to capture this relationship. We
assess the accuracy of the approximation against simulations and and
other well-known simple and straightforward techniques for a wide range
of parameters. Results show that our proposed technique is a
considerable improvement to other standard approximation techniques and
is at the same time not cumbersome.