The phenomenal growth of Internet traffic is attributed to the increasing a
cceptance of the Internet protocol (IP) as the predominant protocol in ente
rprise networks, The traditional design, of a router upon which the Interne
t is based is becoming unable to support the current demand from users. Thi
s service has worked exceptionally well for non real-time traffic such as e
lectronic mail but for real-time traffic such as voice and video, IP has on
ly performed well across a lightly loaded network. To provide services for
real-time traffic, new classes of services art. being introduced into the I
nternet.
For networks to be able to support this increased user demand for more band
width and for multimedia and real-time traffic, IP routers need to be faste
r and to provide quality-of-service (QoS) guarantees. This demand for more
bandwidth and QoS support has fuelled interest in the use of asynchronous t
ransfer mode (ATM) as an underlying link-layer technology in the Internet.
The aim is to exploit the potential benefits of ATM while maintaining the i
nherent strengths of IP.
Hop-by-hop routing normally works well for non-delay sensitive data traffic
but, in cases where prioritisation is required, it gives unacceptable perf
ormance since the current Internet is unable to differentiate between diffe
rent sorts of packets, and thus stipulate that the network provide designat
ed levels of servicing. Since voice and video are delay sensitive they will
add additional pressure to the routers, Additionally the increase in traff
ic is creating the need for higher bandwidth which the packet-forwarding mo
del of router processing ;cannot provide effectively. The goal would be to
expedite movement of information without forcing users to abandon router-in
dependent protocols. Ideally, one would want to move the forwarding functio
n of routing all the way out to the periphery of the network, leaving a pro
tocol-independent network core that would be focused on providing:reliable
high-performance standards-based connections.
The emergence of the multiprotocol label switching (MPLS) architecture(1) a
ims to improve the scalability and performance of the prevalent hop-by-hop
routing and forwarding across packet networks. Its primary goal is to stand
ardise a technology that integrates the label swapping forwarding paradigm
with network layer routing. This label swapping is expected to improve the
flexibility and robustness in. delivering new routing services without alte
ration to the forwarding paradigm. The word multiprotocol suggests that thi
s architecture will be applicable to any network-layer protocol In addition
to IP; label switching indicates the underlying forwarding mechanism. This
paper will start with a basic tutorial on IP switching techniques and then
explain how it can be utilised to accelerate IP packets through the core o
f the network. The paper concludes with the description of a proposed archi
tecture, quality-of-sen,ice label distribution protocol (QoSLDP) whose obje
ctive is to provide bandwidth guarantees.