The explosive growth of Internet users, the increased user demand for bandwidth, and the declining cost of technology, has all resulted in the emergence of new classes of high-speed distributed IP router architectures with packet forwarding rates on the order of gigabits or even terabits per second. This paper develops an analytical framework for modeling and analyzing the impact of technological factors on the cost-performance trade-offs in distributed router architectures. The main trade-off in a distributed router results naturally from moving the main packet forwarding and processing power from a centralized forwarding engine to an ensemble of smaller forwarding engines either dedicated to or shared among the line cards. Processing packets in these smaller engines can be much cheaper (by as much two to three orders of magnitude) than in a centralized forwarding engine. Therefore, the main goal of our modeling framework is to determine an optimal allocation of processing power to the forwarding engines (in a distributed router) to minimize overall router cost while achieving a given level of packet forwarding performance. Two types of router models are analyzed using the proposed framework: a distributed router architecture and parallel router architecture.
[1]
H. M. Ahmed,et al.
IP switching for scalable IP services
,
1997
.
[2]
David Waitzman,et al.
A 50-Gb/s IP router
,
1998,
TNET.
[3]
Srinivasan Keshav,et al.
Issues and trends in router design
,
1998,
IEEE Commun. Mag..
[4]
T. V. Lakshman,et al.
Beyond best effort: router architectures for the differentiated services of tomorrow's Internet
,
1998,
IEEE Commun. Mag..
[5]
H. V. Jagadish,et al.
Towards a Gigabit IP Router
,
1992,
J. High Speed Networks.
[6]
Greg Minshall,et al.
Ip Switching: Atm under Ip *
,
1998
.