Elastic MPLS

RATIONALE AND BENEFITS

COMBINING STATIC AND DYNAMIC MPLS

Overview
The metro’s unique large-scale and simple-topology characteristics pose special challenges for service
providers who wish to evolve to packet. We start with a description of the drivers that led to the emergence
of Carrier Ethernet in general, and MPLS in particular, as viable packet transport technologies. Following this
is a brief review of the differences between IP/MPLS and MPLS-TP flavors. Finally, we present the benefits
and rationale of using Elastic MPLS (both IP/MPLS and MPLS-TP) in the Metro, to address its unique
characteristics.

Streamline L2 and L3 VPN Facilitate service Optimize service

service provisioning, QoS, and OAM visibility for different performance and cost
across MPLS domains and control over large-scale metro profiles
From native Ethernet to carrier Ethernet
Early-generation Ethernet switches supported high-bandwidth Ethernet access at a low cost. However, they
suffered from a number of weaknesses, rooted in the fact that native Ethernet’s scheme of learning bridges
was not intended for use in carrier networks and that the connectionless nature of the technology does not
permit deterministic behavior.
One of the biggest drawbacks of a connectionless technology such as native Ethernet is that it provides no
means of reserving bandwidth or making devices along a path deliver a certain level of service at a specific
bit rate. This means that the aggregation of many services results in unpredictable QoS. In native Ethernet,
traffic is only differentiated by using per-hop behavior based on the values of IEEE 802.1p bits. While 802.1p
offers up to eight levels of priority, it does not provide any guarantees for end-to-end service performance.
Without the ability to control how the traffic flows throughout the network, native Ethernet cannot isolate
guaranteed services for various performance parameters, such as frame delay (latency), frame delay variation
(jitter), frame loss, and resiliency.
Enterprise-centric native Ethernet switches also suffer from other known issues involving: slow restoration,
poor reliability, limited scalability and lack of service management.
MEF (Metro Ethernet Forum) defined a new class of Ethernet, Carrier Ethernet, to comply with the following
five attributes:

Standardized services Service management

enables coordination of subscribers, service enables service providers to roll out, maintain,
providers, and operators, to achieve Carrier and troubleshoot Carrier Ethernet based data-
Ethernet-based data connectivity between multiple connectivity services in a cost-effective and timely
subscriber sites across multiple operator networks. manner.
This has become a necessity for organizations all
over the globe. Standardized services gives service Quality of service
providers guaranteed interoperability between any enables a single network to run multiple services
network elements that are MEF certified. to multiple end-users running a wide variety of
applications with different bandwidth and latency
Scalability requirements - using Carrier Ethernet. It also
enables data connectivity of any number of multiple provides the required tools to ensure that services
end-user sites over any distance. It also supports maintain the performance requirements, according
a wide range of interface speeds for varied traffic to Service Level Specifications (SLS).
demands, allowing ample network growth.

Reliability
enables end-users to employ Carrier Ethernet to run
their business and mission critical applications. These
services run on transport layers that provide the
most stringent resiliency and recovery constraints.
The emergence of MPLS (IP/MPLS)
Standardized by the IETF, MPLS (Multi Protocol Label Switching) is a scalable, protocol-agnostic mechanism
designed to carry circuit and packet traffic over virtual circuits known as Label Switched Paths (LSPs).
Operating at an OSI layer, situated between the traditional definitions of Layer 2 and Layer 3, MPLS makes
packet-forwarding decisions based on the contents of the label without examining the packet itself.
MPLS was developed originally to facilitate packet forwarding by using label switching. But it has additional
attributes, like connection establishment, improved network resiliency, and OAM functions, which help
overcome native Ethernet transport gaps and achieve Carrier Ethernet grade.
MPLS enhances Ethernet’s capabilities across the entire collection of MEF Carrier Ethernet defined attributes,
as follows:

Standardized services Fast Reroute (FRR), for instance, enables service

restoration with SONET/SDH-like protection
MPLS enables a full range of high-performance (<50 ms). Also, MPLS can reduce congestion and
MEF-defined point-to-point (E-LINE), point-to- improve network predictability by enabling a high
multipoint (E-TREE), and multipoint-to-multipoint degree of load balancing – the equal distribution
(E-LAN) connectivity services. In addition, circuit of traffic across LSPs.
emulation services can be delivered using TDM
pseudowires in an MPLS network, ensuring support Scalability
for legacy services. The abundance of MPLS
By layering multiple services onto each MPLS
throughout the Service Provider edge and metro-
LSP, MPLS can support a high number of service
core networks ensures MPLS-based Ethernet
instances in each network. Each service can be
services to extend between metro markets and
provisioned with the bandwidth tailored to the
even across continents. The addition of multicast
particular needs of the customer at each site.
capabilities in MPLS also initiated the possibility of
MPLS-enabled services can scale geographically
efficient IPTV service delivery over a carrier-grade
because they can be provisioned over multiple
infrastructure.
MPLS-based carrier networks. New options, such as
pseudowire switching, help scale Ethernet services
Quality of Service over several networks.
MPLS can transport all types of traffic with
guaranteed QoS per application, regardless Service management
of the protocol of the encapsulated data. LSPs
A number of MPLS OAM tools have emerged in
provide a way to engineer network traffic patterns
recent years that enable operators to better monitor
independent of routing tables, steer traffic flows
network performance, diagnose and isolate faults,
from congested links to alternate links, and control
and centrally manage networks via standard-based
the paths of specific traffic flows to guarantee fast
vendor implementations in line with the MEF’s vision.
resiliency or other QoS parameters, such as frame
New tools also help speed up service provisioning
delay, jitter, or packet loss. MPLS enables bandwidth
and facilitate automation and self-healing. The
to be reserved for selected traffic and admission
MPLS OAM tools complement the Ethernet link
control.
OAM (IEEE 802.3ah) and service management
OAM (EEE 802.1ag and ITU-T Y.1731) tools that
Reliability have been developed over the last years. Together,
MPLS has a variety of local and path-protection these create a multilayer OAM solution to better
mechanisms that help minimize packet loss in an support the service provider requirements for
LSP and ensure high-service reliability. MPLS network and application assurance.
Extending MPLS to the metro
Consumer demands for mobile, cloud, and video services require a lot more bandwidth, forcing telecommunication
service providers worldwide to upgrade existing networking infrastructure with the appropriate technology and
equipment. Metro-only traffic (traffic that traverses only the metro and bypasses long-haul traffic links) already
exceeds long-haul traffic and therefore, the metro infrastructure is crucial for supporting Next Generation
telecommunication services. While IP is the leading technology for communication services, the transport
infrastructure comprises both L2 and L3 VPNs to fit various end-user requirements and preferences.
A key element for cost-efficient metro infrastructure, resides in the MPLS control plane which provides
carrier-grade attributes to the services (QoS, OAM, resiliency). The natural inclination is to extend IP/MPLS
from the core, where it is the de-facto standard, into the metro. However, there are quite a few differences
between the metro and the core, which may affect the optimal choice.

What makes the metro a special case?

Two key attributes make the metro a special case - scale and topology:

Large scale
The core comprises a relatively small number of network elements. In contrast, the metro consists of a large
number of network elements. This number increases as we get closer to the customer premises. As a result, the
NEs used in the metro should be optimized for cost, size, and power consumption and are, therefore, limited
in their processing capability. The large number of NEs also makes simple operation an absolute necessity for
cost-effective service delivery.

Simple topologies
The core uses mesh topology, which provides comprehensive connectivity options. However, the metro is based
on simple topologies like hub and spoke, or ring. As a result, fast restoration schemes that are not based on
complex protocols (and therefore don’t require heavy processing power) are a key concern in metro networks.
The rise of MPLS-TP
Similar to the way MPLS enables the use of Ethernet for transport, MPLS Transport Profile (MPLS-TP)
enables MPLS to be used for transport, and specifically, for metro transport. It is a simplified version of
IP/MPLS, optimized for transport networks. MPLS-TP is both a subset and an extension of IP/MPLS. Some
of the complex IP/MPLS functionality, which is not relevant for transport networks, is disabled while other
transport features are added.
MPLS-TP objectives are to streamline operational models, simplify converged networks, and gain better
control of Ethernet networks in order to enable packet-based services with a similar degree of predictability as
in existing transport networks. To meet these objectives, MPLS-TP is strictly connection-oriented and does
not rely on IP forwarding or routing.
Key benefits offered by MPLS-TP include:

Deterministic transport Improved resiliency

While MPLS-TP and IP/MPLS use the same Predetermined alternative paths ensure sub-50
data-plane mechanism, MPLS-TP uses Bidirectional msec switch to protection.
PseudoWire (PW) and Label Switching Path
(LSP), while IP/MPLS uses Unidirectional PW Lower TCO
and LSP. Having bidirectional traffic following Metro networks are usually large-scale networks.
exactly the same path ensures intuitive, predictable, Thus, the Total Cost of Ownership (TCO) is a
and deterministic transport and bears special significant factor. IP/MPLS usually requires the
importance for providing accurate, reliable packet use of full-blown L3 routers, resulting in higher
synchronization. NE cost (CAPEX) as well as higher operational
expenditures and complexity (OPEX). In contrast,
Full control MPLS-TP can be implemented over simpler devices
Centralized NMS and static configuration provide that do not require distributed control planes and
a broader view of the network and enable the use are easier to operate.
of intelligent planning and prediction tools. These
ensure continuous traffic flow, even in the event of
failure in the control plane.

Transport class OAM

MPLS-TP supports extensive Operation,
Administration, and Maintenance (OAM)
functions similar to those available in traditional
optical transport networks (like SONET/SDH
and OTN). These include enhanced fault
management, performance monitoring, and in-band
PW/LSP/Section OAM levels. The OAM functions
are an integral part of the MPLS-TP data plane and
are independent of the control plane.
Elastic MPLS – the best of both worlds
As explained in the previous sections, both IP/MPLS and MPLS-TP offer advantages to the metro network domain. ECI
sees MPLS technology as a key building block in realizing the efficiency of packet-based transport while maintaining
transport-grade resiliency and manageability. Our view is that MPLS-TP is often a better fit for metro networks,
while IP/MPLS is a better fit for core networks. However, IP/MPLS is the de-facto standard for core networks and is
often a better fit for specific applications and networks in the metro. Therefore, to remove the barriers for extending
MPLS into the metro and match various scenarios and customer preferences, we believe that supporting both
IP/MPLS and MPLS-TP is key to facilitate MPLS penetration in the metro.
The Elastic (Dual Stack) MPLS enables seamless interworking between IP/MPLS and MPLS-TP domains.
This eliminates the need to choose one standard over the other and makes it easier and risk-free to extend
MPLS to the metro. Consequently, service providers can support the dual stack across numerous platforms,
from the metro access and up to the metro core. This gives them the flexibility to set the extent of the
IP/MPLS reach, and switch to either static MPLS (MPLS-TP) or Ethernet.

In addition to standard L3 VPNs (using E2E IP/MPLS) and L2 VPNs (either using static MPLS-TP or dynamic
IP/MPLS), Elastic MPLS offer additional advantages, including:

Lower TCO L3 VPN

Lower TCO is achieved by using MPLS-TP network elements to transport the service in parts of the metro
domains or in its entirety. In addition, it provides a proactive OAM of the working path and the protection path
in the MPLS-TP domain. Even though the L3 VPN service crosses different MPLS domains, Elastic MPLS
provides a unified management plane for end-to-end service and fault correlation.

Streamlined L2 service interworking

Elastic MPLS enables E2E L2 VPN service with unified management plane for E2E service provisioning,
visibility, and fault correlation. This includes E2E OAM across different network domains. Without elastic
MPLS capabilities, service providers would be forced deliver the service via two or more separate services,
per network domain.
Easier control
Elastic MPLS enables you to divide the metro into segments (multisegment pseudowire) and thereby gain
large-scale control more easily, while maintaining an E2E service experience.

True convergence
One of the key ambitions for next-generation transport is running all types of services on a unified networking
layer. This convergence includes both packet-based NG services and TDM-based legacy services. Convergence
should apply to a wide range of applications (mobile backhaul, VPNs, residential triple play, and so on) that
prefer different control planes. An elastic MPLS implementation makes total migration to a unified packet
transport network both a doable and practical goal. Moreover, a single network management system that can
handle both IP/MPLS and MPLS-TP with full operational convergence is a valuable asset.

Risk free
With benefits and sweet spots unique to both IP/MPLS and MPLS-TP, it is most likely that both will be used in
the same network and even within the same network region. With elastic MPLS implementation, the transition
between IP/MPLS and MPLS-TP is inherent and does not require product replacement, hardware upgrades,
or extra training. Thus, dual-stack implementation is risk free.

Optimized service mapping

The flexibility to select IP/MPLS or MPLS-TP per LSP allows network operators to serve different applications
and/or regions with the technology that offers the best fit. For example, while legacy and mission-critical services
are typically mapped to MPLS-TP, residential triple-play and enterprise L3 VPN services are typically mapped
to IP/MPLS. Such flexibility lowers the total cost, simplifies network management, and provides subscribers
with the required quality of service.
Summary
MPLS is supported across ECI’s NPT portfolio, from metro access to metro core, in assorted combinations.

Copyright © 2015 ECI. All rights reserved. Information in this document is subject to change without notice. ECI assumes no responsibility for any errors that may appear in this document.
The various MPLS options allow tailoring the best solution to fit customers’ specific needs, including:
• MPLS-TP
• MPLS-TP and native TDM
• Elastic MPLS (MPLS-TP and IP/MPLS) including a gateway between the two
The rationale for designing and deploying elastic MPLS includes a wide range of commercial, technical, and
economic benefits. With elastic MPLS implementations, service providers enjoy reduced risks from
introduction of new technologies, simpler migration and interoperability, and the lowest possible TCO.
Ultimately, this fulfills the network vision of unified Packet Transport Networks (PTN).

Contact us to find out how our ELASTIC networks can help you grow
ABOUT ECI
ECI is a global provider of ELASTIC network solutions to CSPs, utilities as well as data center operators. Along with
its long-standing, industry-proven packet-optical transport, ECI offers a variety of SDN/NFV applications, end-to-end
network management, a comprehensive cyber security solution, and a range of professional services. ECI's ELASTIC
solutions ensure open, future-proof, and secure communications. With ECI, customers have the luxury of choosing a
network that can be tailor-made to their needs today – while being flexible enough to evolve with the changing needs
of tomorrow. For more information, visit us at www.eci tel e.com