Inter-domain routing is a much more compli-
cated process. Since it deals with routing be-
tween different domains, standardisation is nec-
essary. The most popular protocol today is BGP-
4, which was standardised in 1995. BGP-4 in-
corporates many nice features compared to its
predecessor, the EGP. However, the protocol is
quite complicated, and the whole process relies
heavily on manual configuration of routing
parameters in “BGP configuration table”. This
requires very skilled operators.
With the explosion of Internet traffic, the influ-
ence of the market on the technology develop-
ment has become more pronounced from the
mid 90s and onwards. How to improve network
performance has been a hot topic among the
vendors. Along with the development of hard-
ware technology, research on new routing strate-
gies gained more attention in recent years. The
phenomenal growth of the Internet has also led
to an increased demand on the network to offer
differentiated services. Constraint-based routing
seems to be able to support this by setting differ-
ent constraints and requirements for different
classes of services. Path oriented technologies
such as MPLS have made constraint-based rout-
ing feasible and attractive in public IP networks.
The growth of the Internet leads to more com-
plexity in the routing process; while the emer-
gence of new applications also leads to new ser-
vice requirements. In order to support the new
demands, we believe that the research on routing
will be focused on these areas:
- Inter-domain routing. As the Internet contin-
ues to grow, the number of ASs and the sizes
of individual ASs will increase. BGP will
probably continue to evolve and new para-
meters will be defined. Consequently, the
complexity of inter-domain routing will also
increase. A new routing technology that gives
better support to inter-domain routing will be
needed in the near future. - Multicast routing. The importance of multi-
cast routing is recognised. Examples of ser-
vices that require multicast are conferencing,
streaming audio and video, and interactive
gaming. These applications and services can-
not scale to thousands or millions of receivers
with multiple point-to-point, unicast streams. - Support for mobility. In the long run, it may
well be the case that all computers are mobile.
Even so, they will have to be connected to the
Internet. Therefore, the Internet Protocol must
be extended to support mobility. - Supporting QoS requirements. New appli-
cations, such as multimedia or real-time data,
are sensitive to delay, and even more sensitive
to delay variations (jitter). There are still many
challenges concerning how to solve the rout-
ing problem in a network where traffic with
QoS requirements coexists with best-effort
traffic. We will have to offer QoS that satis-
fies the user requirements while trying to opti-
mise the utilisation of network resources.- Support for IP over Optical Networks. Due
to the high bandwidth requirements between
the IP routers in the future Internet, we expect
that the communication between IP routers in
the core network will be implemented by a
supporting connection oriented circuit
switched network. The strongest candidates
are Optical Transport Network (OTN) and
SDH based on Packet over SDH (PoS). Thus
the IP backbone network will consist of two
networks, the connection-less IP network and
a connection oriented network. Currently there
are heavy research activities on the co-opera-
tion between the IP network and the OTN/PoS
(by IETF, EURESCOM, etc.).
References
[BELL57] Bellman, R E. 1957. Dynamic Pro-
gramming.Princeton University Press.[BLAC00] Black, U. 2000. IP Routing Proto-
cols.Upper Saddle River, NJ, Prentice-Hall.[CHEN98] Chen, S, Nahrstedt, K. 1998. An
Overview of Quality of Service Routing for
Next-Generation High-Speed Networks: Prob-
lems and Solutions. IEEE Network,12 (6),
64–79.[CR-LDP] Constraint-based LSP setup using
LDP.Work in progress, Internet Draft <draft-
ietf-mpls-cr-ldp-05.txt>, February 2001.[CR-notes] Notes on Path Computation in Con-
straint-Based Routing.Work in progress, Inter-
net Draft <draft-kompella-te-pathcomp-00.txt>,
July 2000 (obsolete document).[DIJK59] Dijkstra, E W. 1959. A Note on Two
Problems in Connection with Graphs. Numerical
Mathematics.[EURP616] Eurescom. State-of-the-art of Rout-
ing Principles.Eurescom P616, Task 2, PIR 2.3.[Fedyk] Multiple metrics for traffic engineering
with IS-IS and OSPF.Work in progress, Internet
Draft <draft-fedyk-isis-ospf-te-metrics-01.txt>,
November 2000.[FF62] Ford, L R, Fulkerson, D R. 1962. Flows
in Network.Princeton, New Jersey, Princeton
University Press.