The MPLS FAQ - Page 2 of 3
4. MPLS Components
a. What is a Label? Section 3.1 of RFC
3031: "Multiprotocol
Label Switching Architecture" defines a label as follows "A
label is a short, fixed length, locally significant identifier which
is used to identify a FEC. The label which is put on a particular
packet represents the "Forwarding Equivalence Class" to which that
packet is assigned."
The MPLS Label is formatted as follows:
|-20bits Label-|-3bits CoS-|-1bit
Stack-|-8bits TTL-|
The 32-bit MPLS label is located after the Layer 2
header and before the IP header. The MPLS label contains the
following fields:
- The label field (20-bits) carries the actual
value of the MPLS label.
- The CoS field (3-bits) can affect the queuing and
discard algorithms applied to the packet as it is transmitted
through the network.
- The Stack (S) field (1-bit) supports a
hierarchical label stack.
- The TTL (time-to-live) field (8-bits) provides
conventional IP TTL functionality. This is also called a "Shim"
header.
b. What is a Label Switch Path? An LSP is
a specific path traffic path through an MPLS network. An LSP
is provisioned using Label Distribution Protocols (LDPs) such as
RSVP-TE or CR-LDP. Either of these protocols will establish a
path through an MPLS network and will reserve necessary resources to
meet pre-defined service requirements for the data path.
LSPs must be contrasted with traffic trunks.
From RFC 2702: "Requirements for Traffic
Engineering Over MPLS," "A traffic trunk is an aggregation of
traffic flows of the same class which are placed inside a LSP.
It is important, however, to emphasize that there is a fundamental
distinction between a traffic trunk and the path, and indeed the
LSP, through which it traverses. In practice, the terms LSP
and traffic trunk are often used synonymously. The path through
which a trunk traverses can be changed. In this respect,
traffic trunks are similar to virtual circuits in ATM and Frame
Relay networks."
c. What is a Label Distribution Protocol?
A label distribution protocol (LDP) is a specification
which lets a label switch router (LSR) distribute labels to its LDP
peers. When a LSR assigns a label to a forwarding equivalence class
(FEC) it needs to let its relevant peers know of this label and its
meaning and LDP is used for this purpose. Since a set of labels from
the ingress LSR to the egress LSR in an MPLS domain defines a Label
Switched Path (LSP) and since labels are mapping of network layer
routing to the data link layer switched paths, LDP helps in
establishing a LSP by using a set of procedures to distribute the
labels among the LSR peers.
Label Switching Routers (LSRs) use labels to forward
traffic. A fundamental step to Label Switching is that LSRs agree on
the what labels they should use to forward traffic. They come
to this common understanding by using the Label
Distribution
Label Distribution Protocol is a major part of MPLS.
Similar mechanisms for Label exchange existed in vendor
implementations like Ipsilonâs Flow Management Protocol (IFMP),
IBMâs Aggregate Route-based IP Switching (ARIS), and Ciscoâs Tag
Distribution Protocol. LDP and labels are the foundation of Label
Switching.
LDP has the following basic
characteristics:
- It provides an LSR ă discoveryä mechanism to
enable LSR peers to find each other and establish
communication
- It defines four classes of messages: DISCOVERY,
ADJACENCY, LABEL ADVERTISEMENT, and NOTIFICATION messages
- It runs over TCP to provide reliable delivery of
messages (with the exception of DISCOVERY messages
LDP label distribution and assignment may be
performed in several different modes:
- Unsolicited downstream versus
downstream-on-demand label assignment
- Order versus independent LSP control
- Liberal versus conservative label
retention
d. What's the difference between CR-LDP and
RSVP-TE CR-LDP and RSVP-TE are both signaling mechanisms
used to support Traffic Engineering across an MPLS backbone. RSVP is
a QoS signaling protocol that is an IETF standard and has existed
for quite some time. RSVP-TE extends RSVP to support label
distribution and explicit routing while CR-LDP proposed to extend
LDP (designed for hop-by-hop label distribution to support QoS
signaling and explicit routing). MPLS Traffic Engineering tunnels
are not limited to IP route selection procedures and thus will
spread network traffic more uniformly across the backbone taking
advantage of all available links. A signaling protocol is required
to set up these explicit MPLS routes or tunnels.
There are many similarities between CR-LSP and
RSVP-TE for constraint-based routing. The Explicit Route Objects
that are used are extremely similar. Both protocols use ordered
Label Switched Path (LSP) setup procedures. Both protocols include
some QoS information in the signaling messages to enable resource
allocation and LSP establishment to take place automatically.
At the present time CD-LDP development has ended and
RSVP-TE has emerged as the "winner" for traffic engineering
protocols.
e. What is a "Forwarding Equivalency Class"?
Forwarding Equivalency Class (FEC) is a set of packets which
will be forwarded in the same manner (e.g., over the same path with
the same forwarding treatment). Typically packets belonging to the
same FEC will follow the same path in the MPLS domain. While
assigning a packet to an FEC the ingress LSR may look at the IP
header and also some other information such as the interface on
which this packet arrived. The FEC to which a packet is assigned is
identified by a label.
One example of an FEC is a set of unicast packets
whose network layer destination address matches a particular IP
address prefix. A set of multicast packets with the same
source and destination network layer addresses is another example of
an FEC. Yet another example is a set of unicast packets whose
destination addresses match a particular IP address prefix and whose
Type of Service bits are the same
f. How are Label Switch Paths built? A
Label Switch Path (LSP) is a set of LSRs that packets belonging to a
certain FEC travel in order to reach their destination. Since MPLS
allows hierarchy of labels known as label stack, it is possible to
have different LSPs at different levels of labels for a packet to
reach its destination. So more formally, a LSP of a packet with a
label of level m is a set of LSRs that a packet p has to travel at
level m to reach its destination. Please refer to 3.15 of RFC 3031 - Multiprotocol Label Switching
Architecture, for a very formal and complete definition.
As an example: consider the following
scenario
|------|
1
1 |-----| | R1
|--\
/---| R5 | |------| \
2
2
2 /
|-----|
\|------| |-----| |-----|
/ |
R2 |---| R3 |---| R4
|/ |------|
|-----| |-----|
In the above fig. R1 to R5 are the LSRs that a
packet p must travel to reach its destination. The numbers 1 and 2
are label stack depth. R1 and R5 are border gateway routers and R2,
R3 and R4 are the interior gateway routers. For the purpose of label
forwarding R1 and R5 are peers at the border gateway level and R2,
R3, R4 are peers at the interior gateway level. When R1 receives the
packet P with a label that is 1 level deep, it will swap that P's
label with a corresponding label that will be used by R5. Also since
the packet has to travel through R2, R3 and R4, R1 will push a new
label and the label stack level is now 2. So we have two LSPs here.
One is at level 1 from R1 to R5 and the second is at level 2 from R2
to R4.
In order to build a LSP the LSRs make use of the
routing protocols and the routes learned from these protocols. They
may use other protocols such as RSVP but they are not required
to.
Label Switch Path setup can be either ăindependentä
or ăorderedä control. Independent control of LSP establishment is
when label switching is used in support of destination-based
routing. Here each LSR may make an independent decision to assign a
label to an FEC and to advertise that assignment to its
neighbors. This way the establishment of an LSP follows the
convergence of routing almost immediately. In ordered control, label
assignment proceeds in an orderly fashion from one end of an LSP to
the other. LSP setup may be initiated by the ingress or egress end
of the LSP.
It is possible to support both ordered and
independent control without interoperability issues. Both
approaches have their advantages and disadvantages. Please
refer to Davie, B. and Rekhter, Y., "MPLS:
Technology and Applications" for more details.
Label binding may be created locally and remotely;
by downstream or upstream LSRs; they may be created in response to
control- or data-driven traffic; and the binding may be distributed
in a stand-alone protocol (LDP) or piggybacked on an existing one
(i.e. BGP).
g. What is the relationship between MPLS and the
Interior Routing Protocol Interior Gateway Protocols (IGP),
such as OSPF and IS-IS, are used to defined reachability and the
binding/mapping between FEC and next-hop address. MPLS learns
routing information from IGP (e.g., OSPF, IS-IS). Link-state
Interior Gateway Protocol is typically already running on large
Corporations or Service Providers networks There are no changes
required to IGP routing protocols to support MPLS, MPLS-TE, MPLS
QoS, or MPLS-BGP VPNs.
h. What other protocols does MPLS support besides
IP? By definition, Multiprotocol Label Switching supports
multiple protocols. At the Network Layer MPLS supports IPv6,
IPv4, IPX and AppleTalk. At the Link Layer MPLS supports Ethernet,
Token Ring, FDDI, ATM, Frame Relay, and Point-to-Point Links. It can
essentially work with any control protocol other than IP and layer
on top of any link layer protocol. In addition, development efforts
have allowed MPLS to not only work over any data link layer
protocol, but also to natively carry a data link layer protocol over
IP, thus enabling services such as Ethernet over MPLS.
5. MPLS and ATM
a. What are the differences between MPLS and
ATM? MPLS brings the traffic engineering capabilities of ATM
to packet-based network. It works by tagging IP packets with
"labels" that specify a route and priority. It combines the
scalability and flexibility of routing with performance and traffic
management of layer 2 switching. It can run over nearly any
transport medium (ATM, FR, POS, Ethernet...) instead of being tied
to a specific layer-2 encapsulation. As it uses IP for its
addressing, it uses common routing/signaling protocols (OSPF, IS-IS,
RSVP...)
b. Does MPLS replace ATM? MPLS was not
designed to replace ATM but, the practical reality of the dominance
of IP-based protocols coupled with MPLS's inherent flexibility has
led many service providers to migrate their ATM networks to one
based on MPLS.
MPLS can co-exist with ATM switches and eliminate
complexity by mapping IP addressing and routing information directly
into ATM switching tables. The MPLS label-swapping paradigm is the
same mechanism that ATM switches use to forward ATM cells. For
ATM-LSR the label swapping function is performed by the ATM
forwarding component. Label information is carried in the ATM
Header, specifically the VCI and VPI fields. MPLS provides the control component for IP on both the ATM
switches and routers. For ATM switches PNNI, ATM ARP Server, and
NHRP Server are replaced with MPLS for IP services. The ATM
fowarding plane (i.e 53-byte cells) are preserved. PNNI may
still used on ATM switches to provide ATM services for non-MPLS
ports. Therefore, an IP+ATM switch delivers the best of both
worlds; ATM for fast switching and IP protocols for IP services all
in a single switch.
c. What is "Ships in the
night"? Some vendors support the running of MPLS and ATM in
the same device. Generally speaking, these two processes run
separately. A change in an MPLS path has no bearing on ATM
virtual circuits. This practice is commonly referred to "ships
in the night" since the two processes act alone. However, in
some cases, there is some interaction between the two
processes. For example, some vendors support a mechanism
whereby a reservation of resources by a label switch path is
detected by the ATM control mechanism to avoid resource
conflicts.
"Ships in the night" is used as a transitioning
mechanism as networks migrate their ATM control planes to
MPLS. Networks initially preserve ATM for carrying time
sensitive data traffic such as voice and video, and for connecting
to non-MPLS enabled nodes, while concurrently running MPLS to carry
data. Over time there will no longer be a need for separate
ATM flows and therefore networks will only carry MPLS label-based
traffic.
6. MPLS Traffic
Engineering
a. What does MPLS traffic engineering
accomplish? Traffic engineering refers to the process of
selecting the paths chosen by data traffic in order to balance the
traffic load on the various links, routers, and switches in the
network. Traffic engineering is most important in networks where
multiple parallel or alternate paths are available.
A major goal of Internet Traffic Engineering is to
facilitate efficient and reliable network operations while
simultaneously optimizing network resource utilization and traffic
performance.
The goal of TE is to compute a path from one given
node to another (source routing), such that the path does not
violate the constraints (e.g. Bandwidth/administrative
requirements...) and is optimal with respect to some scalar metric.
Once the path is computed, TE (a.k.a. Constraint based routing) is
responsible for establishing and maintaining forwarding state along
such a path.
b. What are the components of MPLS-TE? In
order to support Traffic engineering, besides explicit routing
(source routing), the following components should be
available:
- Ability to compute a path at
the source by taking into account all the constraints. To do
so the source need to have all the information either available
locally or obtained from other routers in the network (e.g.
Network topology)
- Ability to distribute the
information about network topology and attributes associated
with links throughout the network once the path is computed, need
a way to support forwarding along such a path
- Ability to reserve network resources and to
modify link attributes (as the result of certain traffic taking
certain routes)
MPLS TE leverages several foundation
technologies:
- Constraint shortest path
first algorithm used in path calculation. This is a modified
version of the well known SPF algorithm extended to constraints
support
- RSVP extension used to establish the forwarding
state along the path, as well as to reserve resources along the
path
- Link state IGPs with extension (OSPF with Opaque
LSAs, IS-IS with Link State Packets TLV (type, length,
value)) keeping track of topology changes propagation
c. How does MPLS merge traffic flows?
MPLS allows the mapping from IP packet to forwarding
equivalence class (FEC) to be performed only once at the ingress to
an MPLS domain. A FEC is a set of packets that can be handled
equivalently for the purpose of forwarding and thus is suitable for
binding to a single label.
From a forwarding point of view, packets within the
same subset are treated by the LSR in the same way, even if the
packets differ from each other with respect to the information in
the network layer header. The mapping between the information
carried in the network layer header of the packets and the entries
in the forwarding table of the LSR is many to one. That is packets
with different content of their network layer headers could be
mapped into the same FEC. (example of a FEC: set of unicast packets
whose network layer destination address match a particular IP
address prefix...)
d. How are loops prevented in MPLS
networks? Before focusing on MPLS loops prevention,
let's introduce briefly the different loops handling
schemes.
Generally speaking, loop handling can be split into
two categories:
- Loop prevention: provides methods for avoiding
loops before any packets are sent on the path - i.e. Path
Vector
- Loop mitigation (survival+detection): minimize
the negative effects of loopseven though short term transient
loops may be formed. - i.e. Time-To-Live (TTL). If the TTL
reaches 0, then the packet is discarded
- Dynamic routing protocols which converge rapidly
to non-looping paths
As far as loop mitigation is concerned, MPLS labeled
packets may carry a TTL field that operates just like the IP TTL to
enable packets caught in transient loops to be discarded.
However, for certain medium such as ATM and Frame
Relay, where TTL is not available, MPLS will use buffer allocation
as a form of loop mitigation. It is mainly used on ATM switches
which have the ability to limit the amount of switch buffer space
that can be consumed by a single VC.
Another technique for non TTL segment is the hop
count approach: hop count information is carried within the Link
Distribution Protocol messages [3]. It works like a TTL. Hop count
will decrease by 1 for every successful label binding.
A third alternative adopted by MPLS is an optional
loop detection technique called path vector. A path vector contains
a list of the LSRs that label distribution control message has
traversed. Each LSR which propagates a control packet (to either
create or modify an LSP) adds its own identifier to the path vector
list. Loop is detected when an LSR receives a message with a path
vector that contains its own identifier. This technique is also used
by the BGP routing protocol with its AS path attribute.
e. How does MPLS perform failure recovery?
When a link goes down it is important to reroute all trunks that
were routed over this link. Since the path taken by a trunk is
determined by the LSR at the start of the MPLS path (head end),
rerouting has to be performed by the head end LSR. To perform
rerouting, the head end LSR could rely either on the information
provided by IGP or by RSVP/CR-LDP.
However, several MPLS-specific resiliency features
havebeen developed including Fast Re-Route, RAPID, and Bidirectional
Forwarding. See RFC 3469: "Framework for
Multi-Protocol Label Switching (MPLS)-based Recovery" for
additional information.
f. What differences are there in running MPLS in
OSPF versus IS-IS environments? This is not an MPLS question
but an IGP (Interior Gateway Protocol) question. MPLS
extensions, stated in IEFT RFC's, are supported for both OSPF and
IS-IS. MPLS and BGP-VPN real-world deployments have been on
both protocols for some time now. There is
much debate over which IGP is best. This is usually centered
around scalability. The street word is that IS-IS is more
scaleable than OSPF. That is, a single OSPF area can support
150 plus routers and a single IS-IS area can support 500 plus
routers. However, very large IS-IS and OSPF networks have been
deployed.
Ultimately, it is best to first understand the
benefits and disadvantages of each protocol. Then use the
customer / network requirements to choice the IGP which best suites
your needs.
g. Can there be two or more Autonomous Systems
within the same MPLS domain? This is possible only under very
restricted circumstances. Consider the ASBRs of two adjacent ASes.
If either or both ASBRs summarize eBGP routes before distributing
them into their IGP, or if there is any other set-up where the IGP
routes cover a set of FECs which differs from that of the eBGP
routes (and this would almost always be the case), then the ASBRs
cannot forward traffic based on the top-level label. A similar
argument applies to TE tunnels. Some traffic usually will be either
IP forwarded by the ASBR, or forwarded based on a non-top-level
label.
So there would usually be 2-3 MPLS forwarding
domains if there were two ASes: one for each of the two ASes, and
possibly one for the link between the two ASBRs (in the case that
labelled packets instead of IP packets are forwarded between the two
ASBRs).
Also, it's likely that the ASBRs could not be
ATM-LSRs, as ATM-LSRs typically have limited or no capability of
manipulating label stacks or forwarding unlabelled IP
traffic.
Another example (thanks to Robert Raszuk) is with
the multi-provider application of BGP+MPLS VPNs. As
described earlier, there are usually no *top-level* LSPs established
across the two (or more) provider ASes involved, so it can be argued
that:
- The two ASes are separate administrative domains.
However there are some LSPs established across the two ASes, at a
lower level in the label stack. So, it can be argued that
- The two ASes are the same administrative domain,
in so far as the two providers agree to allow lower-level LSPs
to be established across the two ASes
(1) and (2) are both true, which implies that
different definitions of the boundary of the administrative domains
can exist with respect to different levels in the label stack. It is
also (in hindsight) obvious that different MPLS domain boundaries
can exist with respect to different levels of the label
stack.
7. MPLS VPNs
a. How does MPLS enable VPNs? Since MPLS
allows for the creation of "virtual circuits" or tunnels, across an
IP network, it is logical that service providers would look to use
MPLS to provision Virtual Private Network services. Several
standards have been proposed to allow service providers to use MPLS
to provision VPN services that isolate a customers traffic across
the provider's IP network and provide secure end-to-end connectivity
for customer sites.
It should be noted that using MPLS for VPNs simply
provides traffic isolation, much like an ATM or Frame Relay
service. MPLS currently has no mechanism for packet
encryption, so if customer requirements included encryption, some
other method, such as IPsec, would have to be employed. The
best way to think of MPLS VPNs is to consider them the equivalent of
a Frame Relay or ATM virtual circuit.
b. What alternatives are there for implementing
VPNs over MPLS There are multiple proposals for using
MPLS to provision IP-based VPNs. One proposal (MPLS/BGP
VPNs) enabled MPLS-VPNs via extensions to Border Gateway
Protocol (BGP). In this approach, BGP propagates VPN-IPv4
information using the BGP multiprotocol extensions (MP-BGP) for
handling these extended addresses. It propagates reachability
information (VPN-IPv4 addresses) among Edge Label Switch Routers
(Provider Edge router). The reachability information for a
given VPN is propagated only to other members of that VPN. The BGP
multiprotocol extensions identify the valid recipients for VPN
routing information. All the members of the VPN learn routes to
other members.
Another proposal for using MPLS to create IP-VPN's
is based on the idea of maintaining separate routing tables for
various virtual private networks and does not involve
BGP.
Most implementations of Layer 3 MPLS-VPNs are based
on RFC-2547.
c. What is the "Martini Draft'? The
"Martini Draft" actually refers to set of Internet drafts
co-authored by Luca Martini. These drafts define how MPLS can
be used to support Layer 2 transport services such as Ethernet,
Frame Relay and/or ATM. Martini drafts define Layer 2
encapsulation methods, as well as Layer 2 transport signaling
methods.
Many service providers wish to use MPLS to provision
L2-based services to provide an easy migration for the current L2
service customers, while the providers migrate their networks to
MPLS. Service providers can use standards such as Martini
Draft to provide a myriad of services over their MPLS networks, so
customers can simply choose the technology that is best suited to
their environment.
The Psuedo
Wire Emulation Edge-to-Edge (PWE3) working group is currently
developing standards for Layer 2 encapsulation (including
Draft-Martini and other supporting standards). Current working group drafts can be located at www.mplsrc.com/standards.shtml
under the sub-heading "Layer 2 VPNs and Layer 2
Emulation."
d. What is a "Layer 2 VPN" Layer 2 VPNs
are an extension of the work being undertaken in the PWE3 working
group. Layer 2 VPNs allow service providers to provision Layer
2 services such as Frame Relay, ATM and Ethernet between customer
locations over an IP/MPLS backbone. Service providers can thus
provision Layer 2 services over their IP networks, removing the need
to maintain separate IP and Frame Relay/ATM network
infrastructures. This allows service providers to simplify
their networks and reduce operating expenses.
The IETF's "Layer 2
Virtual Private Networks (l2vpn)" working group is currently
defining standards for provisioning Layer 2 VPN services.
Current working group drafts can be located at www.mplsrc.com/standards.shtml
under the sub-heading "Layer 2 VPNs and Layer 2 Emulation."
e. What is a Virtual Private LAN Service
(VPLS)? VPLS refers to a method for using MPLS to create
virtual LAN services based on Ethernet. In this type of
service, all edge devices maintain MAC address tables for all
reachable end nodes, much in the same way as a LAN switch.
VPLS services enable enterprises to provide Ethernet
reachability across geographic distances served by MPLS
services. Several alternatives for enabling VPLS services are
in development by the L2VPN working group. Please refer to
drafts from that working group for additional information.
Also see the Juniper Network's White Paper "VPLS: Scalable Transparent LAN Services."
f. Are MPLS-VPNs secure? Among many
network security professionals, the term "VPN" implies "encrypted"
tunnels across a public network. Since MPLS-VPNs do not
require encryption, there is often concern over the security
implications of using MPLS to tunnel non-encrypted traffic over a
public IP network. There are a couple of points to consider in this
debate:
- MPLS-VPN traffic is isolated by the use of tags,
much in the same way ATM and Frame Relay PVCs are kept isolated in
a public ATM/Frame Relay network. This implies that security
of MPLS-VPNs is equivalent to that of Frame Relay or ATM public
network services. Interception of any of these three types
of traffic would require access to the service provider
network.
- MPLS-VPNs do not prohibit security. If
security is an issue, traffic can be encrypted before it is
encapsulated into MPLS by using a protocol such as IPSec or
SSL.
The debate over MPLS security really comes down
requirements of the customer. Customers comfortable with
carrying their traffic over public ATM or Frame Relay services
should have the same level of comfort with MPLS-VPN services.
Customers requiring additional security should employ encryption in
addition to MPLS.
For additional information, see Miercom's report "Cisco MPLS-VPN: Security Equivalent to Frame
Relay"
8. MPLS Quality of Service
a. What kinds of QoS protocols does MPLS
support? MPLS supports the same QoS capabilities as IP.
These mechanisms are IP Precedence, Committed Access Rate (CAR),
Random Early Detection (RED), Weighted RED, Weighted Fair Queuing
(WFQ), Class-based WFQ, and Priority Queuing. Proprietary and
non-standard QoS mechanisms can also be support but are not
guaranteed to interoperate with other vendors.
Since MPLS also supports reservation of Layer 2
resources, MPLS can deliver finely grained quality of service, much
in the same manner as ATM and Frame Relay.
b. How do I integrate MPLS and
DiffServ DiffServ can support up to 64 classes while the MPLS
shim label supports up to 8 classes. This shim header has a
3-bit field defined ăfor experimental use. This poses a
problem. This Exp field is only 3 bits long, whereas the
Diff-Serv field is 6 bits. There are different scenarios to
work around this problem.
There are two alternatives that
address this problem called Label-LSP and Exp-LSP models. But
they introduce complexity into the architecture. The diffserv
model essentially defines the interpretation of the TOS bits. As
long as the IP precedence bits map to the Exp bits the same
interpretation as the diffserv model can be applied to these bits.
In the case where additional bits are used in the diffserv model,
one can essentially use the label value to interpret the meaning of
the remaining bits. Recognizing that 3 bits are sufficient to
identify the required number of classes, the remaining bits in the
diffserv model are used for identifying the drop priority and these
drop priorities can be mapped into an L-LSP in which case the label
identifies the drop priority while the exp bits identify the Class
that the packet belongs to.
Many Service Provides have or will
add just a few classes. This small enhancement will be hard
enough to provision, manage and sell. This would be an
effective strategy to get to market quickly with a value-added
service.
The followings classes may be more
appropriate for the initial deployment of MPLS QoS:
-
High-priority, low-latency
"Premium" class (Gold Service)
-
Guaranteed-delivery
"Mission-Critical" class (Silver Service)
-
Low-priority "Best-Effort" class
(Bronze Service)
c. How do I integrate MPLS and ATM QoS
? MPLS makes it possible to apply QoS across very large
routed or switched networks because Service Providers can designate
sets of labels that have special meanings, such as service
class. Traditional ATM and Frame Relay networks implement CoS
with point-to-point virtual circuits, but this is not scalable for
IP networks. Placing traffic flows at the edge into service
classes enables providers to engineer and manage classes throughout
the network.
If service providers manage networks based on
service classes, not point-to-point connections, they can
substantially reduce the amount of detail they must track and
increase efficiency without losing functionality. Compared to
per-circuit management, MPLS-enabled CoS provides virtually all of
the benefit with far less complexity. Using MPLS to establish
IP CoS has the added benefit of eliminating per-VC
configuration. The entire network is easier to provision and
engineer. |