You are on page 1of 13

White Paper

MPLS—An introduction to
multiprotocol label switching
Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Basic routing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Traditional routing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
MPLS traffic engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
MPLS components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Label . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Forwarding Equivalence Class (FEC) . . . . . . . . . . . . . . . . . . . . 6
Label Switch Path (LSP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Label Stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Label Distribution Protocol (LDP) . . . . . . . . . . . . . . . . . . . . . . 6
CR-LDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
RSVP-TE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Interior Gateway Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Border Gateway Protocol (BGP) . . . . . . . . . . . . . . . . . . . . . . . 8
Applications of MPLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Application 1: MPLS in a provider network . . . . . . . . . . . . . . 8
Application 2: MPLS over ATM VCs . . . . . . . . . . . . . . . . . . . . . 9
Future applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Online Comments and Suggestions


Click anywhere in this box to visit our online ‘Reader
Response’ form to comment on this document.

Find Out about Our Other Products


Click anywhere in this box to get on the mailing list
for brochures, planning guides & other information.
Introduction A D
With the exponential growth of the S0 S1 S0
Internet over the last few years, tech-
S2 C S0 S0
F
nology has had to adapt constantly to
new demands for increased bandwidth.
In addition, the Internet will continue B S3 S1 E S1
to see dramatic growth due to the ever-
S0 S1 S0
increasing demand for more bandwidth
to the home, with projections of every
household having broadband access in
Figure 1. Traditional routing example
the future.
With over 250 million new users
projected in the next decade and with cumbersome and complicated. The need gateway protocols (IGPs), such as routing
the implementation of Internet protocol for a simpler forwarding method—one information protocol (RIP) and open
version 6 (IPv6), carriers and service with the traffic management features shortest path first (OSPF), or exterior
providers struggle to scale their current and performance of traditional switches gateway protocol (EGPs) such as border
infrastructures for the inevitable demand combined with the forwarding intelli- gateway protocol (BGP). This is done
on their networks. gence of a router—is definitely felt. by referencing the destination Layer 3
addresses against a routing table for a
In order to meet the growing demand All of these needs can be met with
next hop entry. To clarify, each router
for bandwidth, Internet service providers multiprotocol label switching (MPLS),
that a packet traverses must do a route
(ISPs) need higher performance switching because it integrates the key features of
lookup, based on that destination Layer 3
and routing products. Although most both Layer 2 and Layer 3. Most impor-
address in the IP header. This must be
carrier and service provider core tantly, it is not limited to any Layer 2
performed to determine the packet’s
networks run on impressive asynchro- or Layer 3 protocol. In particular,
next hop in its path to get it to its final
nous transfer mode (ATM) backbones, MPLS has several applications and can
destination. The Layer 2 destination
most connections to these providers be extended across multiple product
address is then replaced with the address
continue to be slow frame relay and segments (such as an MPLS router, an
of the next hop’s Layer 2 address, and the
point-to-point connections, introducing IP services switch/router, a multiservice
source Layer 2 address is then replaced
latency and sometimes bottlenecks at switch, an Optical Ethernet switch, as
with the Layer 2 address of the current
the edge access points. Core network well as optical switches).
router, leaving the source and destina-
routers also contribute to latencies, as tion Layer 3 addresses in place for the
each must make its own individual next hop to perform its own route lookup
Basic routing
decision on the best way to forward each on the packet. This process must be
incoming packet. Traditionally, IP has Traditional routing repeated at each hop to deliver the
been routed over ATM using IP over packet to its final destination. In Figure
In traditional routing environments, a
ATM via virtual circuits (VCs) or multi- 1, to forward packets to Router F,
packet is forwarded through a network
protocol over ATM (MPOA). These Router C will reference only the desti-
on a hop-by-hop basis using interior
forwarding methods proved to be
3
nation address of Router F. Router C MPLS traffic engineering become quite apparent. When an LSR
will then determine the best route In contrast, MPLS is a method of or LER constructs its LIB, explicit control
based on the attributes that are defined forwarding packets at a high rate of speed. can be used to direct a packets route
for that particular IGP. If the router is It combines the speed and performance through a network.
using RIP, the lowest sum of all hops to of Layer 2 with the scalability and IP The true strength of the MPLS
the destination is preferred as the best intelligence of Layer 3. Routers on the forwarding algorithm is that analysis of
path, as long as the total number of edge of the network (label edge routers the IP packet header only needs to be
hops does not exceed 15. If the IGP is [LERs]) attach labels to packets based done once, at the ingress of the MPLS
OSPF, the total cumulative cost (i.e., on a forwarding equivalence class (FEC). domain by an LER. Once a packet has
metric, usually based on bandwidth) to Packets would then be forwarded through been assigned to a FEC, forwarding of
a destination is referenced, and the the MPLS network, based on their the packet can be done solely on the
lowest total cost of all links is preferred. associated FECs, through swapping the labels used by the particular label switch
Running IGPs such as RIP and OSPF labels by routers or switches in the core path (LSP).
have provided scalable solutions, but fall of the network called label switch
routers (LSRs) to their destination. In Figure 2, we see that by using MPLS,
short when introducing the need for
we have granular control over a packet’s
inter-AS (autonomous system) routing,
path. Packets destined to Router F
network management, traffic engineering,
The most important idea is that sourcing from Router A follow the solid
and scalable IP services. To reference
green LSP. Packets originating from
Figure 1 again, in traditional routing by adding a simple label, the LSR
Router B will be forwarded along the
environments, Router C must make its is able to switch a packet much dotted red LSP. This is accomplished
forwarding decisions for packets destined more efficiently, due to a simple by referencing incoming labels to the
to Router F based on the metrics defined
by the IGP being implemented. If OSPF,
forwarding element, in contrast LIB in order to attain the value of the
to the hop-by-by basis used in outgoing label and the outgoing inter-
the metric can be based on various
face. Table A is an example of a router’s
criteria, although bandwidth is usually traditional routing.
LIB. Packets destined for a FEC off
the one used. RIP, on the other hand
Router F will arrive at Router C on
uses a metric based on hop count and
interface S2 with a label of 50. They
drops packets after 15 hops. As seen in A label is usually a locally significant,
will be referenced against an LIB to
Figure 1, all packets coming from condensed view of the IP header, which is
ascertain a forwarding decision. Based
Router A or B destined for Router F bound to a FEC (such as a given IP
on the LIB, packets with a label value
will be forwarded in the same way, along prefix) that is then referenced against a
of 50 arriving on interface S2 will be
the path with the preferred metric. table of incoming labels to outgoing
swapped for a label with a value of 12
Therefore, if the path to Router F via labels and interface mappings called
and be switched out interface S0.
Router D is a higher bandwidth—such the label information base (LIB). The
as two DS-3s—and the path via Router label itself represents the particular The strength of controlling traffic
E was connected through T-1s, the path FEC to which it was mapped. By refer- patterns quickly becomes apparent
via Router D would be the only one encing an LIB, the true strengths of when compared to traditional IGPs
used unless a network failure occurred. MPLS traffic engineering capabilities (such as OSPF), which forward packets

4
A D An additional benefit that MPLS provides
S0 S1 S0
is that upgrades to the protocol can be
done easily because the forwarding and
S2 C S0 S0 F control components are separate. The
forwarding component is responsible
S3 S1 S1 for transporting a packet based on a
B E routing table. The control component
S0 S1 S0 is responsible for the construction and
maintenance of the route table, as well
as working with the control components
Figure 2. MPLS traffic engineering example of other nodes to disseminate routing
information. We will discuss each of
these in greater detail as we continue
through the components section.
Table A. Router C’s LIB
Label
The label is a condensed view of the
Interface IN Label In Destination Exit Interface Label Out header of an IP packet, although
contained within it is all of the infor-
S2 50 F S0 12
mation needed to forward the packet
S3 45 F S1 98 from source to destination. Unlike the
IP header, it does not contain an IP
address, but rather a numerical value
agreed upon by two MPLS nodes to
based on the destination Layer 3 function at a higher degree of perform- signify a connection along an LSP. The
address only. Most OSPF network ance and intelligence than current label is a short, fixed-length, physically
designs with multiple paths to the same technologies. It also provides a more contiguous identifier which is used to
destination use only the route with the efficient manner of forwarding packets identify a FEC, usually of local signifi-
lowest accumulative cost. from source to destination than the cance. A packet assigned to a given FEC
hop-by-hop basis used in traditional is usually based on its destination address,
routing, as described earlier. Many of either partially or completely. The label,
MPLS components its components are simply extensions which is put on a particular packet,
MPLS employs many new enhancements of already existing technologies—such represents the FEC to which that packet
to IP routing in the forwarding of as the extensions added to existing is assigned.
packets. Many of these enhancements routing protocols, which will be described
are similar to traffic engineering and in further detail later in this document. Within some transport mediums, there
quality of service (QoS) techniques Additionally, LSR/LER functionality are existing labels that can be used by
employed in ATM. Other components can be added to ATM or optical switches MPLS nodes when making forwarding
of the MPLS protocol enable it to simply by upgrading software. decisions, such as ATM’s virtual path
5
identifier/virtual circuit identifier 0 1 2 3
(VPI/VCI) field and frame relay’s data 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
link connection identifier (DLCI). Other | Label | EXP | S | TTL |
technologies, such as Ethernet and +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-++-+-+-+-+-+-+-+-+-+-+
point-to-point links, must use what is
Label (20)—A locally significant ID used to represent a particular FEC during the forwarding process.
called a shim label, shown in Figure 3.
EXP (3)—Previously called class of service (CoS), it is now considered an experimental range. Currently, this field
is being considered for QoS implementations.
The shim label is a 32-bit, locally signifi-
cant identifier used to identify a FEC. S (1)—Used to signify if label stack is present. If the label is the only one present or at the bottom of the stack, the
bit will be a value of zero.

TTL (8)—Field used to signify the number of MPLS nodes that a packet has traversed to reach its destination. The
Forwarding Equivalence value is copied from the packet header and copied back to the IP packet header when it emerges from the LSP.

Class (FEC)
A FEC can be thought of as any set of Figure 3. Shim label alignments
packets that are forwarded in the same
way through a network. A FEC can
include all packets whose destination topmost label is swapped. The labels are ular network. The LDP was designed
address matches a particular IP organized in a last-in, first-out manner. solely for this use, but LDP alone cannot
network prefix, or packets that belong In other words, the topmost label signi- meet QoS needs. In order to support
to a particular application between a fies the highest LSP, and each successive QoS applications, an LDP must be
source and destination computer. FECs label signifies the next lowest LSP. able to properly select and reserve
are usually built through information network resources along an LSP. In
learned through an IGP, such as OSPF Label Distribution Protocol (LDP) order to support this, one can either
or intermediate system to intermediate In order for an LSR to swap the label use an existing protocol used for
system (IS-IS). on an incoming packet and forward it resource reservation and extend it for
to its downstream peer, it must have a label distribution, or take a protocol
Label Switch Path (LSP) method of learning what label value that can be used for label distribution
its downstream peers are expecting. and extend it to support resource reser-
The LSP is essentially the predetermined
Currently, several protocols can be used vation. An example of a protocol that
route that a set of packets bound to a
for the distribution of labels betweeen already supports this type of reserva-
FEC traverse through an MPLS network
LSR peers, such as LDP, constraint- tion is resource reservation protocol
to reach their destination. Each LSP is
based routed-label distribution protocol (RSVP). An example of distribution
unidirectional; therefore, return traffic
(CR-LDP), resource reservation protocol protocols that can be extended to
must use a separate LSP.
(RSVP), and BGP. reserve, resources are LDP and BGP.

Label Stack The MPLS architecture RFC 3031 The LDP Specification Internet Draft
By placing multiple labels onto a packet, does not specify that any one protocol defines LDP as the set of procedures
MPLS can support a hierarchal routing should be used for the distribution of and messages by which LSRs establish
design. The set of labels attached to a labels between LSR peers. In fact, the LSPs through a network by mapping
packet is called the label stack. As the protocol used should depend on what network-layer routing information
packet traverses the network, only the requirements must be met by a partic- directly to data-link layer switched paths.
6
However, as these LSPs are built, there of ATM. It also has the ability to allocate communicates with two basic types of
is also a need to ensure that they can bandwidth based on an LSP’s priority messages, PATH and RESV. PATH
support CoS and traffic engineering (ranging from 0-7) and/or its age. messages flow from a sender to one or
requirements. In addition, there are some Capabilities of CR-LDP include: multiple receivers. Upon receipt of a
cases where there is a need for a method PATH message, a receiver can send an
1. Remains in a hard state and can be
of creating explicitly routed LSPs through RESV message in return. The label
thought of as a nailed-up connection.
an MPLS domain. As stated before, itself is carried within the RESV message.
LDP alone cannot support this, so it 2. Has both explicit setup and explicit Figure 5 shows an example of a down-
has been extended to support constraint- teardown. stream allocation mapping of labels.
based routing (CR-LDP). CR-LDP is Label distribution flows in the opposite
3. Needs no refresh; once up, it stays
not alone in offering this functionality; direction of the LSP flow. In other
up until torn down.
RSVP is another protocol (more fully words, the label bound to a FEC is
discussed in section 3.4.2) that provides 4. Provides a neighbor discovery mech- received from its downstream neighbor.
many of the same benefits of CR-LDP anism by multicasting hello messages
LSP Path
but with traffic engineering extensions; as a user datagram protocol (UDP) FEC 1

i.e., RSVP-TE (traffic engineering). packet to the LDP port at the “all
The following descriptions are given so routers on this subnet” multicast
that a high level understating can be group address to find all directly RESV, Label 92 RESV, Label 44 RESV, Label 23

gained about how each signaling protocol connected LSRs [1].


performs to attain the same end result. Figure 5. Label request for FEC 1
5. CR-LDP is an extension on an already
A more detailed description of each is via RSVP-TE
existing and running protocol, LDP.
beyond the scope of this document.
What is important is to gain an under- Label Req Label Req Label Req
FEC 1
standing of some of the nuances of each.
RSVP-TE:
CR-LDP 1. Remains in a soft state, requiring
Label Map 92 Label Map 44 Label Map 23
Constraint-based routed LDP is one the retransmission of refresh messages
method of establishing point-to-point in order to maintain an LSP.
Figure 4. Label request for FEC 1
LSPs and QoS in MPLS. These attrib- 2. Provides downstream-on-demand
via CR-LDP
utes are particularly useful when label distribution.
attempting to engineer links over the
public Internet or when setting up a 3. Reserves network resources for
RSVP-TE
virtual private network (VPN). CR-LDP explicit LSPs.
RSVP is another similar method of
allows for explicit routes, using both strict 4. Uses IP datagrams to transport
establishing a point-to-point LSP that
and loose hops, providing maximum messages between peers instead of
meets QoS requirements for MPLS. It
flexibility in building a specific path using TCP and, therefore, must
is an extension of the original RSVP
through a network. Some of the appeal handle the loss of distribution
protocol, with new capabilities to
to CR-LDP is that its version of QoS messages itself.
support an MPLS domain [5]. RSVP
is similar to the rock-solid technology
7
Customer A
Interior Gateway Protocol Border Gateway Protocol (BGP)
ISP 1
IGPs are routing protocols used within BGP is a routing protocol usually used ISP 2

an autonomous system (AS) to provide between autonomous systems (AS) to MPLS


intra-AS routing. Most interior routing provide inter-AS routing. It is currently Network

protocols used today fall under two widely deployed to interconnect large Customer B
different models, distance vector and provider networks into what we call
link state. The details of these models the Internet. In MPLS, BGP can also
are beyond the scope of this document. be used to distribute label-binding
In production environments, link state information for each route it advertises. Figure 6. Provider network
algorithms are usually preferred for This is made possible by the multipro-
their resiliency, scalability, and built-in tocol extensions (MPEs) to BGPv4. A
intelligence. Common link state proto- label for a route can be piggybacked Application 1: MPLS in a
cols are OSPF and IS-IS. For MPLS within the same UPDATE message used provider network
environments, typical link state routing to advertise it to its peer. The label and There are several implementations
protocols are being extended to support any related information is carried as within service provider and carrier
the construction of LSPs that meet part of the network layer reachability networks where MPLS can be used to
specific QoS requirements. An example information (NLRI). offer more services. MPLS can be used
is OSPF-TE, where LSA Type-10 will in conjunction with BGP for scalable
support additional information inter-AS routing, by alleviating the
required for QoS uses. Applications of MPLS
need for internal AS routers to receive
This document presents several possible BGP routes as long as they support
While these extensions are being used
scenarios of MPLS in different networks. MPLS [2]. This allows more control
for traffic engineering, it is important
The goal of the following sections is to over the path in which transit traffic
to note that IGPs are used for providing
provide a high-level overview of some flows through a provider’s network
a next hop label forwarding entry
possibilities of using MPLS. These are through explicitly routed LSPs to mini-
(NHLFE). The NHLFE is used when
not the only ways to use MPLS. In mize underutilized links. This decreases
forwarding a labeled packet by providing
fact, implementations of MPLS are the possible load placed on network
the packet’s next hop, and carries infor-
limited not by the protocol itself, as it resources.
mation on what to do with the packet
can be used to offer circuit emulation,
upon receipt. If the next hop is itself, Another benefit an MPLS implementa-
Internet connectivity, IP/VPNs, and
the LSR will pop the top level label and tion provides is the ability for service
QoS for IP—and much more. We will
“forward” the resulting packet to itself. providers and carriers to provide network-
briefly touch on some applications to
Another forwarding decision will then based VPNs, a service that can give
give examples of how versatile the
be made, based on what remains after companies an alternative to expensive
protocol is.
the label stacked is popped. At this point, private leased line networks with an
the packet may still be labeled, or it may inexpensive solution of “tunnels” through
be a native IP packet. If it is a native a provider’s network emulating the
IP packet, the LER will then make a leased line network. VPNs also conserve
forwarding decision based on its IGP. public IP addresses by allowing private
8
network addresses to traverse the explicitly defined PVCs to be built for offer multi-service features can continue
Internet while encapsulating a private each pair of nodes. NHLFEs can be to provide ATM/FR services as usual
IP address link within a virtual tunnel. provided by private network-to-network while being migrated to an MPLS
interface (PNN) or IGP of choice. This environment.
ATM Network
greatly simplifies future implementations
of MPLS in an ATM environment, as
well as providing a simple migration to UBR Traffic
MPLS in the core of an existing ATM
network.
CBR Traffic
MPLS Domain

Figure 9. MPLS/ATM QoS mapping

Figure 7. IP over ATM overlay


N(N-1)/2 problem ATM Network
MPLS CoS can also be used to imple-
ment ATM’s QoS features, allowing
Application 2: MPLS over
providers to reliably offer voice and
ATM VCs
video services as well as traditional data
In traditional networks, IP routing has transport. Figure 9 shows how voice or
been implemented over ATM networks video traffic can be assigned to a certain
Figure 8. MPLS over an existing
by creating permanent virtual circuits CoS, and then mapped as ATM CBR
ATM network
(PVCs) between IP hosts. This type of traffic. Data traffic can be assigned a
network design can become complex default best-effort CoS (usually 0) and
very quickly when multiple hosts are An added benefit of an MPLS imple- mapped as unspecified bit rate (UBR)
added to the network. For every one mentation over an existing ATM traffic, because it traverses the ATM
host added there would need to be network is that it is not required that network to its final destination. Such
N(N-1)/2 (where N is the total number every device in the MPLS domain be an capabilities will be greatly needed, as
of nodes) PVCs created to provide a LER or LSR. MPLS is can be imple- companies switch to voice over IP (VoIP)
full mesh to all nodes. Figure 7 shows mented in the same network that is from traditional public switched tele-
that if a provider wanted a network of also simultaneously operating standard phone networks (PSTNs) and as more
four routers, they will need to create Layer 2 protocols, also known as “ships- multimedia applications are ported to
six PVCs to keep a full mesh network. in-the-night.” It also does not impose the Internet.
If four more routers are added, four- additional configuration on non-MPLS
teen more PVCs—a total of twenty— switches [2]. This allows more freedom Future applications
will be needed to keep the full mesh when designing and migrating to MPLS As we end our examination of present
environment. With MPLS, IP packets from an existing network infrastructure. MPLS applications, we can begin to
can be routed directly over the data Moreover, ATM switches that currently see where future evolutions and appli-
link plane, eliminating the need for
9
cations of this technology, primarily tion, allowing service providers to take pushing and popping—combined with
generalized multiprotocol label the flexibility of MPLS and apply it to the CoS controls contained in higher
switching (GMPLS) (formerly known an optical (and particularly dense wave- layers—allows it to offer a granularity
as multiprotocol lambda switching length division multiplexing [DWDM]) so far unsurpassed.
[MPLambdaS]), can carry us. As carrier framework. A basic overview of this
MPLS, particularly its applications in the
grade networks continue to increase in technology is as follows: instead of an
optical realm (through GMPLS), creates
size and dependability, they also need LER assigning a label to a particular
a protocol scalable in ways unheard of
to flatten. The ability to collapse their segment of an LSP from its label infor-
until now. The technology powering
Layer 1, 2, and 3 networks onto one mation database, the optical cross
GMPLS is micro-electric mechanical
platform is becoming increasingly connect (OXC) will act as a GMPLS
systems (MEMS). The ability to use
apparent. MPLS offers the ability to LER. The GMPLS LER will assign an
micro-mirrors to redirect lambdas has
dynamically transport any Layer 2 or 3 individual lambda (referred to as a color
opened the doors to a bandwidth
protocol through any MPLS aware or wavelength) to a particular segment
explosion. One of the limitations that
network. GMPLS is the next evolu- from its wavelength forwarding infor-
surfaced with mimetic switching has
mation base. One of the things that will
turned out to be a positive progression
allow this technology to be scalable with
in the end. Until now, we have always
the Internet is the ability of GMPLS to
wanted to see to the bit level within
provision optical paths using the tech-
each box in our network. As of now we
niques embedded in MPLS for traffic
Label, 67 have no way to “see” into light waves
engineering and ISIS/OSPF extensions.
without terminating them in some type
This technology is particularly pertinent
of optical-electrical-optical (OEO) device
now that Layer 2 and 3 networks are
which, while necessary in some places,
Label, 89 moving towards MPLS. As the protocols
adds unneeded latency in others. In
powering all layers of the open systems
Lbl Stack, getting away from terminating lambdas,
Green/89 interconnection (OSI) model begin to
we have had to take a daunting step:
use similar, if not the same, protocols
trusting a box in our network to success-
many of the complex networking prob-
fully switch packets without manage-
lems of the past begin to dissolve.
Lbl Stack, ment past the lightwave level. While
Red/89 Lbl Stack,
Blue/89
Another approach to solving this problem this seems to lack some of the redun-
Label, 89 is using Ethernet over fiber. While this dancy and stability of older technologies,
solution is solid in the local area network it more than makes up for that lack
(LAN) and metropolitan area network with its scalability and speed. It is also
(MAN), its feasibility is limited in the much less of a concern when we have
Label, 52
wide area network (WAN). Carriers fully protected rings and equipment
and other WAN players are looking for that can successfully deliver carrier-
more feature ability than traditional grade reliability. This evolution, while
Figure 10. Example of label stacking Ethernet gives. The tunneling availabil- taken in baby steps, will allow us to
using GMPLS ities of MPLS, as well as the multi-layered eventually have networks that require

10
much less management and overhead, the protocol. Each can be found LDP—Label Distribution Protocol
allowing for more bandwidth and less on the working group’s website at
LER—Label Edge Router
headache. http://www.ietf.org/html.charters/
mpls-charter.html. LIB—Label Information Base

Summary LSP—Label Switch Path


MPLS is the natural evolution for Acronyms LSR—Label Switch Router
existing networks to provide the neces- AS—Autonomous System
MAN—Metropolitan Area Network
sary capabilities to support the explosive
ATM—Asynchronous Transfer Mode
growth of the Internet, while at the same MEMS—Micro-Electric Mechanical
time enabling network administrators BGP—Border Gateway Protocol System
to control traffic at a more granular
CoS—Class of Service MPE—Multiprotocol Extension
level. The act of simply swapping a
label instead of referencing an IP header CR-LDP—Constraint-based Routing MPLambdaS—Multiprotocol Lambda
at each hop provides a more efficient Label Distribution Protocol Switching
manner of forwarding packets, which in
DLCI—Data Link Connection Identifier MPLS—Multiprotocol Label
turn allows the opportunity for traffic to
Switching
be forwarded at tremendous speeds. DWDM—Dense Wavelength Division
Additionally the multiprotocol part of Multiplexing MPOA—Multiprotocol over ATM
MPLS allows flexibility on the type of
EGP—Exterior Gateway Protocol NHLFE—Next Hop Label Forwarding
media used as a transport, be it ATM,
Entry
frame relay, gigabit Ethernet, or Packet FEC—Forwarding Equivalence Class
over SONET (PoS)—allowing providers NLRI—Network Layer Reachability
FTN—FEC to NHLFE Map
and carriers tremendous opportunities Information
for expansion. GMPLS—Generalized Multiprotocol
OEO—Optical-Electrical-Optical
Label Switching
The protocol is now an official standard
OSI—Open Systems Interconnection
with the tremendous need for its deploy- GRE—Generic Route Encapsulation
ment. Further information on MPLS can OSPF—Open Shortest Path First
IETF—Internet Engineering Task Force
be located at the Internet Engineering
OXC—Optical Cross-Connect
Task Force (IETF) website, where an IGP—Interior Gateway Protocol
established MPLS working group is PNNI—Private Network-to-Network
ILM—Incoming Label Map
developing the protocol for standardi- Interface
zation. The IETF working group IP—Internet Protocol
PoS—Packet over SONET
currently has several Internet drafts that
ISP—Internet Service Provider
can provide a deeper understanding PSTN—Public Switched Telephone
of the protocol. “Multiprotocol Label L2—Layer 2 Network
Switching Architecture” and “A Frame-
L3—Layer 3 PVC—Permanent Virtual Circuit
work for MPLS” are the best starting
points for a deeper understanding of LAN—Local Area Network
11
QoS—Quality of Service References
RIP—Routing Information Protocol 1 Andersson et al., “Label Distribution
Protocol Specification,” work in
RSVP—Resource Reservation Protocol progress (draft-ietf-mpls-ldp-11),
SVC—Switched Virtual Circuit February 2001.

SVP—Switched Virtual Path 2 Callon et al., “Framework for


Multiprotocol Label Switching,”
TCP—Transmission Control Protocol work in progress (draft-ietf-mpls-
TE—Traffic Engineering framework-05), March 2000.

TTL—Time-To-Live 3 Rosen et al., “Multiprotocol Label


Switching Architecture,” work in
UBR—Unspecified Bit Rate progress (draft-ietf-mpls-arch-07),
UDP—User Datagram Protocol January 2001.

VC—Virtual Circuit 4 D. Awduche et al., “Requirements


for Traffic Engineering Over MPLS,”
VCI—Virtual Circuit Identifier RFC 2702, September 1999.
VoIP—Voice over IP 5 D. Awduche et al., “Applicability
VP—Virtual Path Statement for Extensions to RSVP
FOR LSP-Tunnels,” work in
VPI—Virtual Path Identifier progress (draft-ietf-mpls-rsvp-tunnel-
VPN—Virtual Private Network applicability-01), October 2000.

WAN—Wide Area Network

12
In the United States: In Canada: In Europe: In Asia: In Australia:
Nortel Networks Nortel Networks Limited Nortel Networks Nortel Networks Nortel Networks
4006 Hwy. 54 8200 Dixie Road Maidenhead Office Pk. Singapore Pre Ltd Australia Pty Ltd
P.O. Box 13010 Suite 100 Westacott Way 151 Lorong Chuan 380 St. Kilda Rd.
RTP, NC 27709 Brampton, Ontario Maidenhead Berkshire #02-01 5th/6th Fl.
L6T 5P6 SL6 3QH New Tech Park Melbourne, Victoria
Tel: +44 1628 432 000 Singapore, 556741 3004
Fax: +44 1628 437 666 Tel: 65 287-2877 Tel: 613 9206 4646

Published by:
For more information, contact your Nortel Networks representative. Call 1-800-4 NORTEL
(1-800-466-7835) in North America or 1-506-674-5470 outside North America. Nortel Networks
Marketing Publications
http://www.nortelnetworks.com
© Copyright 2001 Nortel Networks Corporation. Printed in USA, April 2001. Information subject to change. Nortel Networks
Dept. 0526
Corporation reserves the right, without notice, to make changes in equipment design or components as changes in engineering or
manufacturing methods warrant. Nortel Networks and the globemark are trademarks of Nortel Networks Corporation.
P.O. Box 13010
55053.25/04-01 Issue 2 • Printed in USA April 2001 RTP, NC 27709

You might also like