1. Carrier Ethernet Services -
The Future
Public Multi-Vendor
Interoperability Test
Berlin, September 2008
2. Carrier Ethernet World Congress 2008 Multi-Vendor Interoperability Test
EDITOR’S NOTE INTRODUCTION
This year the interoperability This year’s interoperability event focused on the
hot staging test for the Future of Carrier Ethernet Services. While each
Carrier Ethernet World previous event concentrated on specific topics such
Congress took place in as mobile backhaul or service creation, this event
parallel to the Beijing aimed to congregate the knowledge and experience
Olympics. the industry gained in the last four years into a single
80 engineers from 28 parti- modern, converged network showing all that a tier-
cipating vendors with over one service provider is likely to encounter. We
therefore tested:
Carsten Rossenhoevel 100 systems attended our
Managing Director test. According to data from • Converged residential, business and Mobile
Heavy Reading, more than Backhaul services
90% of the Carrier Ethernet switch and router market • Clock synchronization
share were represented in this test.
• Business services realized using E-Line, E-LAN
The participating vendors verified 34 test areas in and for the first time E-Tree services
any-to-any combinations in ten days, truly
challenging the Olympic motto “Faster, Higher, • The leading access, transport and aggregation
technologies
Stronger“. Carrier Ethernet implementations support
more functions and cover more markets today — • Microwave access and transport
ranging from core to microwave to access, E-Lines to • Ethernet OAM: Fault management and perfor-
E-Trees, triple play to mobile backhaul. mance monitoring
It was an outstanding experience to witness the • High availability
massive testing feast, a unique get-together of
• Management and SLA reporting
virtually all leading players with
one single goal: To improve In order to construct such a large test
multi-vendor interoperability of network and cover all the above test
advanced Carrier Ethernet imple- TABLE OF CONTENTS areas a ten day, closed doors hot
mentations. staging event was conducted at
Participants and Devices ..............3 EANTC’s lab in Berlin, Germany.
An EANTC panel of service
providers worldwide including Network Design..........................4 Since the first Carrier Ethernet
2 experts from COLT, GVT Brazil, Interoperability Test Results ...........4 World Congress in 2005, EANTC
PT TELKOM Indonesia, T-Systems has organized interoperability test
Ethernet Service Types .................4
and Metanoia Inc reviewed the events which are then showcases at
Diverse Access Technologies ........6 the congress.
test plan thoroughly to ensure the
event’s scenarios are realistic Diverse Transport ........................7 Our interoperability showcases are
and sound. MPLS Core .................................8 driven by three main goals:
Interestingly, market forces are E-NNI ........................................9 Technical – Through participation in
operating at full strength. This Mobile Backhaul.......................10 the event, vendors have the oppor-
year, we once again tested three tunity to verify the interoperability of
Clock Synchronization ...............14
transport technologies in the test their devices and protocol imple-
event’s metro/aggregation Ethernet OAM ..........................15 mentations against the majority of
networks: MPLS, PBB-TE and Resilience and Fault Detection.....18 the industry’s leading vendors.
T-MPLS. These three compete to Management and SLA Reporting.21 Marketing – The participants can
some extent — at our test, they all showcase the interoperability of
Acronyms.................................22
proved being well suited for the their latest solutions on a unique,
transport of Carrier Ethernet References ...............................23
large-scale platform.
services.
Standards – When fundamental
Service OAM support is becoming mandatory for issues are found during the hot staging event EANTC
aggregation and CPE devices; the Ethernet reports the discoveries to the standard bodies. These
microwave market flourishes; mobile backhaul in turn update the standards.
pushes support for backwards compatibility (ATM
EANTC started the preparation for the event by
pseudowires, circuit emulation) and new features
inviting interested vendors to weekly conference
(clock synchronization, IEEE 1588v2, E-Tree, among
calls during which the technical and marketing goals
others).
for the event were discussed and agreed. The test
This white paper summarizes in detail the plan, created by EANTC based on the test topics
monumental effort that the participating vendors and suggested by the vendors, expanded on the
EANTC team underwent. Enjoy the read. experience gained from previous events and was
lined up with recent IEEE, IETF, ITU-T and MEF
standards.
3. Participants and Devices
PARTICIPANTS AND DEVICES Vendor Participating Devices
Vendor Participating Devices Nortel Metro Ethernet Routing
Switch (MERS) 8600
Actelis Networks ML658
RAD Data ACE-3205
ADVA Optical FSP 150CC-825 Communications ACE-3200
Networking ACE-3400
Alcatel-Lucent 1850 TSS-40 ASMi-54
5650 CPAM Egate-100
7450 ESS-6 ETX-202A
7705 SAR ETX-202A/MiRICi
7750 SR7 ETX-202A/MiTOP
9500 MPR IPMUX-216/24
LA-210
Calnex Solutions Paragon Sync OP-1551
RICi-16
Cambridge VectaStar
RICi-155GE
Broadband Networks
Redback Networks — SmartEdge 400
Ceragon Networks FibeAir IP-MAX2 an Ericsson Company
FibeAir IP-10
Rohde & Schwarz SIT SITLine ETH
Ciena LE-311v
LE-3300 SIAE ALS
MICROELETTRONICA ALFO
Cisco Systems 7606
7604 Spirent Spirent TestCenter
ME4500 Communications
Catalyst 3750-ME
Symmetricom TimeProvider 5000 PTP
ME-3400-2CS
Grand Master 3
ME-3400-12CS TimeCesium 4000
ECI Telecom SR9705 Tejas Networks TJ2030
Ericsson Marconi OMS 2400 Telco Systems — T5C-XG
a BATM Company T5C-24F
Foundry Networks NetIron XMR 8000
T5C-24G
Harris Stratex Eclipse (Gigabit) Radio T-Marc-250
Networks T-Marc-254
T-Marc-340
Huawei Technologies NE5000E Cluster System
T-Marc-380
NE40E-4
T-Metro-200
CX600-4
Tellabs 8830 Multiservice Router
InfoVista VistaInsight for Networks
Ixia XM2 IxNetwork
Juniper Networks M10i
MX240 Service Provider Test Plan Review
MX480 The draft test plan was reviewed by a panel of
global service providers in July this year. Their
NEC Corporation CX2600 feedback and comments were reflected in the final
PASOLINK NEO version of the test plan. EANTC and the partici-
PASOLINK NEO TE pating vendors would like to thank: COLT, GVT
Brazil, PT TELKOM Indonesia, T-Systems and
Nokia Siemens hiD 6650 Metanoia Inc.
Networks Flexi WCDMA BTS
FlexiHybrid
RACEL
4. Carrier Ethernet World Congress 2008 Multi-Vendor Interoperability Test
NETWORK DESIGN INTEROPERABILITY TEST RESULTS
As in previous events we set off to construct a In the next sections of the white paper we describe
network that would allow all participating vendors to the test areas and results of the interoperability
establish end-to-end Ethernet services with any of the event. The document generally follows the structure
other vendors. One of the central design consider- of the test plan.
ations for the network was to enable any device Please note that we use the term »tested« when
positioned in the access network to reach any other reporting on multi-vendor interoperability tests. The
access network device regardless of the other term »demonstrated« refers to scenarios where a
device’s point of attachment to the network. This service or protocol was terminated by equipment
proved to be especially useful for such end-to-end from a single vendor on both ends.
tests as Service OAM or Mobile Backhaul. The
specifics of these tests can be found in the test case
sections. ETHERNET SERVICE TYPES
We also aimed to build a network that would look
The Metro Ethernet Forum (MEF) has defined three
familiar to service providers. It is perhaps unrealistic
Ethernet service types in order to allow the industry
to expect that service providers will incorporate all
and specifically the customers interested in the
current transport technologies into their network.
services to have a common language to discuss such
Nevertheless the familiar network domains are likely
Ethernet based services. The three service types are
to exist: access, aggregation, metro and core,
defined in terms of the Ethernet Virtual Connection
regardless of the chosen transport technology. It is
(EVC) construct:
realistic, however, to expect service providers to use
MPLS in the core. • E-Line – Point-to-point EVC
Looking at the network from a customer’s • E-LAN – Multipoint-to-multipoint EVC
perspective, we used the following network areas: • E-Tree – Rooted-multipoint EVC
• Access: The devices that normally exist at the While the E-Line service type provides a service to
customer premise or by NodeBs or base stations exactly two customer sites, the E-LAN and E-Tree
were positioned here. We were lucky to see a service types allow the connection of more than two
diverse number of access technologies for trans- customer sites. In contrast to the E-LAN service type
porting Ethernet such as microwave links, which allows an any-to-any connectivity between
copper, and fiber. These devices implemented customer sites, E-Tree introduces two different roles
4 the UNI-C construct as defined by the MEF. for customer sites: leaf and root. An E-Tree service
• Aggregation: The aggregation area of a network facilitates communication between leaves and roots,
consisted of a variety of solutions meant to however, leaves can not communicate with each
aggregate customer premise devices. This other directly. An E-Tree service implemented by a
included Provider Bridges and H-VPLS Multi- rooted-multipoint EVC can be used to provide
Tenant Unit Switches (MTU-s). When applicable multicast traffic distribution and hub-and-spoke
these devices performed the UNI-N role in the topologies (e.g. DSL customers to BRAS).
network.
In the test network we instantiated three specific
• Metro: Three different transport technologies definitions of service types: Ethernet Virtual Private
were used in each of the three metro area Line (EVPL), Ethernet Virtual Private LAN (EVP-LAN),
networks: MPLS, PBB-TE and T-MPLS. This and Ethernet Virtual Private Tree (EVP-Tree). All
allowed each transport technology to test its own services were configured manually in the network.
resiliency and Network-to-Network Interface
Due to the increasingly large amount of devices and
(NNI) solutions.
vendors we had present at the hot staging, this
• Core: As stated above, IP/MPLS was used to process was time consuming and prone to mistakes.
support connectivity between the different metro A multi-vendor provisioning tool would have been
area networks in order to realize end-to-end ideal for the testing and is recommended for any
services. In addition, MPLS Layer 3 VPNs as service provider planning to deploy Carrier Ethernet
defined in RFC 4364 were tested in the core of services.
the network.
The services created in the network were configured
The physical network topology presented here
in two ways:
depicts the roles of all the devices and their
respective placement in the network. Please note that • EVCs that remained within the same metro area
many tests required logical connectivity between the network
devices, often at an end-to-end nature, which will be • EVCs that crossed the network core
shown, where applicable, using logical topologies The sections below describe the services in the
in each test section. network in detail.
5. Ethernet Service Types
E-Tree core, two of which provided E-NNI leaf connectivity
to the other metros - the Alcatel-Lucent 7750 SR7 to
For the first time at an EANTC interoperability event, T-MPLS and the ECI SR9705 to PBB-TE. The Tejas
an E-Tree service instantiation was established. One TJ2030 interpreted this E-NNI connection as the root
EVP-Tree was configured with one root node within connectivity for the PBB-TE metro, and the Ericsson
the MPLS metro area and leaves throughout all Marconi OMS 2400 did the same for the T-MPLS
network areas. The MEF defines an E-Tree service to metro.
be a rooted Ethernet service where the roots are
able to communicate with all leaves, and all leaves The diagram in figure 1 shows all points where
are able to communicate with the roots, but not with E-Tree traffic was verified. The logical connections
each other. represent something different in each area: Ethernet
pseudowires in the MPLS, PBB-TE trunks in the
This service utilized each metro technology in a PBB-TE, and TMCs in the T-MPLS networks.
unique way. The MPLS metro used a separate VPLS
instance to create this service, using different split
horizon groups to ensure that leaf UNIs could only E-LAN
communicate with the root UNI, but could not
establish communication between each other. The One EVP-LAN was configured in the network with
Cisco ME4500 implemented the root UNI-N and customer ports in all three metro areas. The
handed the service off to the Nokia Siemens construction of the EVP-LAN service used different
Networks hiD 6650 which propagated the tree into mechanisms in each metro area. These mechanisms
the MPLS metro. The Juniper MX480 configured a are described in details in the diverse transport
leaf using MPLS towards the Cisco 7606 which section.
treated this connection as the root for the core
network. Three leaves were configured within the
Cisco ME-3400-2CS
Cisco ME4500
Telco Systems
T-Marc-340 Nokia Siemens Networks
hiD 6650
MPLS 5
ECI
SR9705
Telco Systems
MTU-s
MTU-s T-Metro-200
Telco Systems Huawei Juniper Redback Tellabs
T-Metro-200 CX600-4 MX480 SmartEdge 8830
400
MTU-s MTU-s
Cisco Juniper
7604 MX480 Telco Systems Ciena
Cisco T-Metro-200 LE-311v
Ceragon 7606 ADVA
FibeAir IP-10 FSP 150CC-825
Ericsson Alcatel-Lucent ECI
7750 SR7 SR9705 Tejas TJ2030
Marconi OMS
2400
Foundry
Ericsson NetIron XMR 8000 Tejas TJ2030
Marconi OMS T-MPLS PBB-TE
2400
Alcatel-Lucent
1850 TSS-40 Nortel Ciena
Ericsson
MERS 8600 LE-3300
Marconi OMS
Telco Systems
2400
T-Marc-380
ADVA Cambridge Harris Stratex
FSP 150CC-825 VectaStar Eclipse
Access MTU Root UNI Leaf UNI
Device MTU-s Switch
Aggregation Metro/Core Root/Leaf Logical Path
Device Device E-NNI Propagation
Figure 1: E-Tree logical connections
6. Carrier Ethernet World Congress 2008 Multi-Vendor Interoperability Test
E-Line DIVERSE ACCESS TECHNOLOGIES
The E-Line service type configured in the network
The different services in the test network used a
used Virtual LAN (VLAN) IDs to distinguish between
variety of access technologies to reach the simulated
the various services. In some cases, much like real
last mile customer access device. Most services used
world networks, a switch positioned at the customer
fiber (multi-mode) and copper based Gigabit
site would add a Service VLAN tag (S-VLAN) to the
Ethernet. One UNI was implemented over a single
Ethernet traffic provided by the customer, therefore,
strand fiber cable using IEEE 802.3ah defined
allowing the customer to maintain its private VLAN
1000BASE-BX10 between the Cisco ME4500 and
addressing scheme and separate the customer VLAN
the Cisco ME-3400-2CS. Two Actelis ML658
space from the provider’s.
devices used G.SHDSL.bis to connect the aggre-
gation area to the access. RAD demonstrated a wide
variety of access technologies including EFM
bonding of four G.SHDSL.bis pairs between the
ASMi-54 and LA-210. In addition, RAD demon-
strated Ethernet over PDH connectivity with the
ETX-202A with MiRICi E1/T1 over a single E1 link
T-MPLS and the RICi-16 over 16 bonded E1 links, both
aggregated by the Egate-100. The PDH to
channelized STM-1 was performed by the OP-1551.
MPLS PBB-TE Several Ethernet access links comprised of two
Ethernet links with a microwave signal in between.
These systems are described in more detail below.
Microwave for Access and
Transport
In recent years we have seen an increased interest in
our interoperability events from vendors offering
UNI-C Microwave Access Device microwave connectivity and network solutions.
UNI-C Access Device Microwave solutions alleviate the need to roll out
6 physical wire infrastructure and are especially
UNI-N Aggregation Device prevalent in such areas as cellular backhaul,
emerging markets, large corporation networks,
UNI-N Metro Device
hospitals, and mobile-fixed operators.
User Network Interface (UNI) This event enjoyed the participation of the following
microwave products: Alcatel-Lucent 9500 MPR,
Figure 2: E-Line service creation Cambridge VectaStar, Ceragon FibeAir IP-10 and
FibeAir IP-MAX2, Harris Stratex Eclipse, NEC
All vendor devices successfully participated in PASOLINK NEO, Nokia Siemens Networks Flexi-
creation of E-Line services. From the number of Hybrid, and SIAE MICROELETTRONICA ALS and
combinations tested, we are confident that an any-to- ALFO. In addition, Cambridge Broadband Networks
any combination of endpoints is possible. provided a point-to-multipoint microwave system
Three of the E-Line services created between the with which providers can connect either multiple
three metro clouds were encrypted using Rohde & customer offices or multiple base stations via
Schwarz SITLine ETH. The encryption device was Ethernet or E1 lines.
situated between the UNI-C (which was emulated by Since the radios rely on a signal through the air
Spirent TestCenter) and Alcatel-Lucent 7705 SAR, some weather events such as rain and heavy fog
Telco Systems T-Marc-380 and Telco Systems can cause the signal to degrade effectively
T-Metro-200 all of which were serving as UNI-N decreasing the range or capacity of the link. Radio
devices. Once the encryption connections were devices can recognize the decrease in air-link
established we verified that the EVCs were indeed capacity and some solutions can distinguish which
encrypted and that the connection remained stable. frames should be prioritized and further transported
versus which frames will be dropped. The Alcatel-
Lucent 9500 MPR, Cambridge VectaStar, Ceragon
FibeAir IP-10, Harris Statex Eclipse, and SIAE
MICROELETTRONICA ALFO showed this function-
ality by decreasing the modulation scheme in
Quadrature Amplitude Modulation (QAM) which
caused traffic loss only to best effort frames and no
or minimal loss to prioritized traffic with unaffected
latency.
7. Diverse Transport
Services relying on microwave equipment will need of a single Virtual Private LAN Services (VPLS)
to be made aware when the microwave signal is too instance utilizing both VPLS PEs and H-VPLS MTU
weak to transmit traffic. The link state propagation switches established between the following devices:
function disables the Ethernet link state for all ports Alcatel-Lucent 7450 ESS-6, Ciena LE-311v, Cisco
associated with the microwave link. This function- 7604 and Catalyst 3750-ME, ECI SR9705, Huawei
ality was demonstrated by the Ceragon FibeAir CX600-4, Ixia XM2 IxNetwork, Juniper MX240 and
IP-10, Harris Stratex Eclipse and the SIAE MX480, Nokia Siemens Networks hiD 6650,
MICROELETTRONICA ALFO. These devices also Redback SmartEdge 400, Tellabs 8830, and Telco
showed their capability to propagate an incoming Systems T-Metro-200. This VPLS instance used LDP
loss of signal on a tributary Ethernet port across the for signaling statically configured peers as
microwave link and switching off the appropriate described in RFC 4762. These devices also estab-
physical port on the other side of the radio lished Ethernet pseudowires using LDP to facilitate
connection. point-to-point Ethernet services.
Cambridge Broadband Networks demonstrated A separate VPLS instance was used to test BGP-
their ability to share point-to-multipoint link capacity based Auto-Discovery, which was successfully estab-
between several end stations. In the demonstration lished between the Cisco 7606 and the ECI
three end stations were defined to share a 45 Mbps SR9705. A total of four vendors were interested in
wireless link to a central controller. Cambridge testing BGP-based Auto-Discovery, one of which
Broadband Networks showed that when capacity uncovered an interoperability issue during the tests
between one base station and controller was not where packets captures were taken to be further
used, the remaining base stations could use the studied in their labs.
extra capacity. In order to test the interoperability of VPLS implemen-
Over the last few years we have seen an impressive tations which use BGP for signaling as described in
increase in the features built into microwave RFC 4761, another separate VPLS instance was
transport. While historically microwave solutions configured. This was tested between the following
were used to provide a virtual wire, we see more devices with BGP-based Auto-Discovery enabled:
and more intelligence built into the solutions — on Huawei CX600-4 and Huawei NE40E-4, and
several products a complete Ethernet switch Juniper MX240 and Juniper MX480. The Juniper
functionality. MX480 performed an interworking function
between this BGP signaled VPLS domain and an LDP
signaled VPLS domain with the Cisco 7604.
DIVERSE TRANSPORT
7
The Carrier Ethernet architecture specified by the Provider Backbone Bridge
MEF is agnostic to the underlying technology used to
Traffic Engineering (PBB-TE)
provide Carrier Ethernet services. The creation and
support of such services is, however, an essential One of the potential solutions to delivering MEF
component of the interoperability test event. Mainly defined services using Ethernet technologies only is
three technologies compete for Carrier Ethernet the IEEE defined Provider Backbone Bridge Traffic
Transport: MPLS, PBB-TE and T-MPLS. During this Engineering (PBB-TE). The technical specification is
event we had the opportunity to verify all three defined in 802.1Qay which is working its way
technologies. The following sections describe test through the standard process and is in draft version
results for each technology in detail. 3.0 at the time of the testing. The standard extends
the functionality of the Provider Backbone Bridges
(802.1ah) adding a connection-oriented forwarding
MPLS mode by creating point-to-point trunks. These trunks
MPLS is defined in a set of protocols standardized deliver resiliency mechanisms and a configurable
by the Internet Engineering Task Force (IETF) and the level of performance.
IP/MPLS Forum. MPLS is positioned to deliver layer The vendors participating in the PBB-TE transport
2 and layer 3 services including Ethernet services as domain included Ciena LE-311v2, Ciena LE-3300,
defined by the MEF while being agnostic to the Ixia XM2 IxNetwork, Nortel MERS 8600, and Tejas
underlying transport technology. TJ2030.
The tests in this area were based on previous In the PBB-TE Metro network we were able to test the
experience gained from EANTC’s Carrier Ethernet establishment of E-Line, E-LAN, and E-Tree services.
World Congress and MPLS World Congress interop- The establishment of E-Line services was straight-
erability test events and reached a larger number of forward as we tested it in several previous events.
participants than in previous events, including a total E-LAN and E-Tree services creation was tested tested
of 12 vendors testing MPLS implementations. The for the first time within the PBB-TE cloud. For the
MPLS metro domain operated independently from E-LAN service Ciena LE-3300 and Tejas TJ2030
the MPLS core network. switches established bridging instances per PBB-TE
The MPLS metro network was built solely for the trunk and C-VLAN/S-VLAN IDs. Every PBB-TE edge
purpose of Carrier Ethernet services. Multipoint-to- device established a trunk for each particular UNI to
multipoint services were facilitated with the creation one of the bridges. A few issues related to usage of
different Ethertype values in CFM messages,
8. Carrier Ethernet World Congress 2008 Multi-Vendor Interoperability Test
padding, and different interpretation of CCM using a multipoint architecture similar to VPLS. On
intervals were discovered in the initial configuration one particular E-Line service, both Alcatel-Lucent
phase of PBB-TE trunks, however, these issues were 1850 TSS-40 and Ericsson Marconi OMS 2400
resolved quickly. were able to successfully test Quality of Service
In addition, the Nortel MERS 8600 and the Spirent (QoS) by distinguishing between three different
TestCenter tested one of the latest additions to classes of service within the same Ethernet service
Ethernet - Provider Link State Bridging (PLSB) – a pre- and only drop low priority traffic when interfaces
standard implementation of the IEEE 802.1aq were oversubscribed.
(Shortest Path Bridging) which is in draft version 0.3. Since the T-MPLS standards do not define a control
The protocol uses the IETF defined IS-IS protocol for plane protocol, the T-MPLS connections between
distributing Backbone MAC addresses and Service vendors were manually configured. Ericsson used
IDs of participating nodes across the network. Once two proprietary management tools (ENEA and
the network topology has been learned, IS-IS is used DiToNe) to setup the T-MPLS network configuration
to establish loop-free multipoint-to-multipoint services. on their devices.
The forwarding plane uses PBB (802.1ah), however
since the other devices in the PBB-TE network did not
support PLSB the three Nortel MERS 8600 devices
T-MPLS to MPLS-TP Migration
were able to use a re-encapsulation of either PBB-TE Following the approval of the first version of the
trunks or VLAN tags to peer within the PBB-TE ITU recommendations on T-MPLS, the IETF and
network. The Nortel MERS 8600 devices and the ITU-T jointly agreed to work together to extend
Spirent TestCenter emulated nodes successfully MPLS protocols to meet transport network
learned the appropriate B-MAC addresses, and requirements in order to ensure a smooth
forwarded the respective traffic accordingly. convergence of MPLS-based packet transport
Tejas Networks demonstrated a logical Ethernet LAN technology. A Joint Working Group (JWT) was
network with an IEEE 802.1ad based Ethernet Ring formed between the IETF and the ITU to achieve
Protection Switching (ERPS). This ring based control mutual alignment of requirements and protocols
protocol being standardized under ITU-T G.8032 is and to analyse the different options for T-MPLS
a protection mechanism which offers carriers a standard progress. On the basis of the JWT
deterministic sub-50 ms network convergence on a activity, it was agreed that the future standard-
fiber failure as opposed to the conventional loop- ization work will focus on defining a transport
breaking mechanisms like Rapid Spanning Tree profile of MPLS (named MPLS-TP) within IETF
Protocol (RSTP). ERPS convergence time is and in parallel aligning the existing T-MPLS
8
independent to the number of nodes in the network, Recommendations within ITU-T to the MPLS-TP
thereby vastly enhancing the scalability of a carrier work in IETF.
network. We measured failover and restoration of At their Dublin meeting in July 2008, the IETF
below 35 ms for the demonstrated ERPS. has initiated the work on MPLS-TP. Due to the
fact that IETF MPLS-TP standard or drafts do not
exist yet, we tested the implementations based
Transport MPLS (T-MPLS) on the T-MPLS ITU-T Recommendations currently
This test marked the third T-MPLS interoperability in force and its relevant drafts.
testing at EANTC. The following devices successfully It is our intention also to include the first imple-
participated in the T-MPLS area during the event: mentations of MPLS-TP drafts at our next event.
Alcatel-Lucent TSS-40, Ericsson Marconi OMS
2400, and Ixia XM2 IxNetwork.
The T-MPLS standards specify the networking layer
for packet transport networks based on MPLS data
MPLS CORE
plane and designed for providing SONET/SDH-like Since MPLS is used by the majority of service
OAM and resiliency for packet transport networks. providers as core technology it is only logical that
Alcatel-Lucent and Ericsson successfully tested the when providers add Carrier Ethernet services to their
creation of E-Line, E-LAN and E-Tree services, the last product offering the MPLS core will be used. We
of which was a first at an EANTC interoperability followed this approach and used the MPLS core to
event. Both participants constructed T-MPLS paths connect between three different Ethernet transport
(TMP) which are end-to-end tunnels that aggregate metro areas. The core area was constructed using
T-MPLS channels (TMC) representing the services. the following edge devices, all of which successfully
The TMPs and TMCs were transported over different established multiple VPLS domains and CVirtual
physical layer types including 1 Gbit Ethernet, Private Wire Services (VPWS) using LDP for various
10 Gbit Ethernet, ITU-T G.709, and SDH STM-16. Ethernet services: Alcatel-Lucent 7750 SR7, Cisco
The Alcatel-Lucent 7705 SAR was used as a non- 7606, ECI SR9705, Foundry NetIron XMR 8000,
T-MPLS switch in the aggregation area, interfacing to Huawei NE40E-4, Juniper M10i, and Tellabs 8830.
the T-MPLS domain by means of statically configured All devices were physically connected to Huawei
MPLS labels. NE5000E cluster system P router (P for Provider, as
The E-LAN and E-Tree services were configured opposed to PE for Provider Edge) through which all
services were tunneled through by default.
9. External Network to Network Interface (E-NNI)
In addition to providing transport for Ethernet MPLS Metro Connectivity to the Core
services, all edge devices in the core established an
IP/MPLS L3VPN service using BGP (based on RFC Several options exist to allow connectivity between
4364). The Alcatel-Lucent 7750 SR7 and Cisco two MPLS areas. The preferred options were MPLS
7606 terminated Ethernet pseudowires into this VPN based, but one option used IEEE 802.1ad Provider
providing the potential to offer layer 3 services to Bridging tags, or simply 802.1Q VLAN tags to
customers which are not reachable otherwise. transport services between the two areas. The Label
Edge Router (LER) in the MPLS core would strip the
MPLS header from traffic before it forwarded the
EXTERNAL NETWORK TO Ethernet frames to the LER in the MPLS metro. The
S-Tag or VLAN tag would then signal to the MPLS
NETWORK INTERFACE (E-NNI) metro device which pseudowire to forward the
T-MPLS PBB-TE frames onto. Devices using VLAN tags were Alcatel-
Tejas Lucent 7450 ESS-6, ECI SR9705, and Foundry
Alcatel-Lucent Ciena
TJ2030 NetIron XMR 8000.
Ericsson 1850 TSS-40 LE-3300
Marconi The other option used to connect between adminis-
OMS 2400 tratively separated MPLS core and metros is referred
to as pseudowire (PW) stitching. This involves the
creation of two pseudowires, one in each domain,
Alcatel-Lucent ECI Nortel and then interconnecting them either within one
7750 SR7 SR9705 MERS 8600
device, or with a third pseudowire between the two
edge devices. Vendors who took this approach
Huawei Tellabs 8830 chose the latter. In this case two MPLS labels must be
NE40E-4 signaled: the inner label (PW label) signaled by
Foundry LDP, and the transport label (PSN label) signaled by
Juniper
NetIron XMR 8000
M10i either eBGP (IPv4+label) or LDP. To facilitate the
Cisco transmission of LDP sessions, either a separate OSPF
7606
area was enabled between the two edge devices or
a static route was used. Devices taking the PW
stitching approach were Cisco 7606, Juniper M10i,
Juniper MPLS Alcatel-Lucent Juniper MX480, and Redback SmartEdge 400.
MX480 7450 ESS-6
Redback ECI
SmartEdge 400 9
SR9705
PBB-TE Connectivity to the Core
Metro/Core Device As PBB-TE and 802.1ad are both part of the IEEE
Provider Bridging domain of technologies, it is not
Provider Bridging (802.1ad)
surprising that Provider Bridging tags were
LDP-based E-NNI supported across the board in the PBB-TE metro
eBGP (IPv4+label)-based E-NNI domain. All services crossing the core into the
VLAN (802.1Q) PBB-TE cloud used S-Tags (Service Tags) to distin-
Static MPLS Configuration guish each service between a core edge router and
a PBB-TE switch. These devices included Ciena
Figure 3: E-NNI to the core LE-3300, ECI SR9705, Nortel MERS 8600, Tejas
TJ2030, and Tellabs 8830. One PBB-TE trunk was
configured between Nortel MERS 8600 and Tejas
As we described above three different technologies
TJ2030 and traversed the MPLS core.
were used in the metro areas. The problem that
every service provider then faces is to connect the
metro area with the existing network core. In our test T-MPLS Connectivity to the Core
network, much like in most service provider
networks, the core used MPLS for transport and Two options were used to establish services over an
services. Therefore, we required mechanisms to MPLS core into a T-MPLS network. The first was to
allow services originating on one metro area to use 802.1ad S-Tags, similarly to the MPLS metro.
cross the core and be received on other metro The second option used was pseudowire stitching.
areas. The T-MPLS edge device terminated TMCs coming
The following subsections describe the specific from the edges of the network and stitched them to
Network to Network Interface (E-NNI) solutions used an MPLS Ethernet pseudowire which was estab-
in the network. lished with the neighboring core edge device. This
was done using LDP for both MPLS labels, which ran
over a separate OSPF area than the core. This
option was tested between Alcatel-Lucent 1850
TSS-40 and the Alcatel-Lucent 7750 SR7, which also
had some services configured over statically
configured MPLS pseudowires.
10. Carrier Ethernet World Congress 2008 Multi-Vendor Interoperability Test
MOBILE BACKHAUL Carrier Ethernet
Network
Traditionally, the interface between mobile base Requirements
stations and base station controllers has been based
on a number of parallel TDM circuits (for GSM, Mobile Network
TDM Circuit Emulation
Sub-second Resiliency
Sub 50-ms Resiliency
CDMA) or ATM connections (for the first versions of Service Types
ATM Pseudowires
UMTS and CDMA 2000) carried on E1 or T1 links.
E-Lines Services
E-LAN Services
E-Tree Services
Several market studies show that the transport
network costs account for 20–30% of a mobile
operator’s operational expenditure (OPEX).
SGSN
BSC Traditional GSM X X
MSC
Mobile Core
Traditional 3G (UMTS X X
Backhaul Network Rel. 99 / CDMA2000)
RNC
Hybrid 3G offload X X X
GGSN
Radio Access Network (ATM-based voice; IP
tunneled data)
MSC — Mobile Service Switching Center Full packet-based 3G a X X
SGSN — Serving GPRS Support Node
GGSN — Gateway GPRS Support Node Long-Term Evolution X X
BSC — Base Station Controller (LTE, 4G)
RNC — Radio Network Controller
Mobile WiMAX a X X
Figure 4: Mobile Backhaul scope
a. Can be used as a fallback
With the advent of high-speed data transport (HSPA)
in 3G networks, with WiMAX and LTE on the Circuit Emulation Services (CES)
horizon, the amount of data traffic in mobile
10 There is a number of specifications defining circuit
networks has vastly grown and will continue to do
so. Mobile operators are considering mobile emulation services which could be used to support
backhaul over Carrier Ethernet networks, as these Mobile Backhaul services. During our event we
provide enough bandwidth for any predicted tested implementations and observed demonstrations
increase in data traffic and are more cost effective of MEF8 and RFC 4553 specification for E1 inter-
faces.
than the current TDM networks.
During the tests and demonstrations the devices were
The main issue and test focus for Mobile Backhaul
connected either back-to-back, back-to-back with an
transport is the migration path from TDM/ATM to
impairment generator of Calnex Paragon Sync
converged packet based services. Thousands of
emulating a network behavior between the two
base stations will not be upgraded immediately or
devices under test, or over the whole test network.
not at all. Migration paths vary widely depending
We accepted a test or a demonstration if the two
on the specific service provider environment.
devices performing circuit emulation were able to
In this test event, we verified a number of migration pass E1 data over the packet based network and the
scenarios, focusing TDM and ATM transport over deviation of the E1 signal received from the network
Carrier Ethernet as well as clock synchronization. compared to its input signal was within 50 parts per
The following table provides an overview of the million (ppm) over 10 minutes.
transport requirements imposed by different mobile As shown in the CES tests back-to-back figure in total
network technologies. 5 products from three different vendors passed the
tests: Alcatel-Lucent 9500 MPR, NEC CX2600, NEC
PASOLINK NEO TE, RAD IPMUX-216/24, and RAD
MiTOP-E1/T1 hosted by the RAD ETX-202A. All tests
were performed using MEF8 for encapsulation and
adaptive clocking for clock synchronization. The test
between Alcatel-Lucent 9500 MPR and RAD
IPMUX-216/24 was performed once with and once
without the optional RTP (Real Time Protocol) header.
All other tests were performed without RTP. As
described in RFC 4197 section 4.3.3, the usage of
RTP relaxes the tolerance requirement for the internal
clocks of the devices performing CES and therefore
11. Mobile Backhaul
decreases the probability of jitter buffer overflow or Alcatel-Lucent demonstrated MEF8 CES with differ-
underflow. ential clocking by using RTP (Real Time Protocol)
In addition, CES was demonstrated over the whole header between Alcatel-Lucent 9500 MPR and
test network as shown in the diagram ., and tested Alcatel-Lucent 9500 MPR devices over T-MPLS metro
back-to-back with the Calnex Peragon Sync network. In the same demonstration Alcatel-Lucent
impairment tool, as shown in Figure 6: CES tests showed its proprietary solution to synchronize two
across the network. The Calnex Paragon Sync microwave endpoints over the air by transporting
emulated the jitter of a network path with 10 nodes the clock information from the Alcatel-Lucent 9500
and 40% traffic utilization for a more realistic MPR performing the CES to another Alcatel-Lucent
scenario. 9500 MPR over the airFigure 6: CES Across the
Network.
.
Alcatel-Lucent NEC PASOLINK Alcatel-Lucent Alcatel-Lucent
9500 MPR NEO TE 9500 MPR 9500 MPR
Alcatel-Lucent RAD RAD ACE-3400 RAD ACE-3205
9500 MPR IPMUX-216/24
Alcatel-Lucent Calnex RAD ETX202A
RAD ACE-3200 RAD ACE-3205
9500 MPR Paragon Sync with MiTOP E1/T1
Alcatel-Lucent Calnex NEC CX2600
9500 MPR Paragon Sync NEC CX2600 NEC CX2600
RAD Calnex Alcatel-Lucent
9500 MPR Access network TDM service
IPMUX-216/24 Paragon Sync
UNI TDM link
RAD NEC CX2600
IPMUX-216/24 11
Metro network Access Device
TDM service
TDM link Figure 6: CES tests across the network
Access Device
Figure 5: CES tests back-to-back ATM Transport over MPLS
As described in the introduction, ATM transport
A number of microwave solutions participated in this services over a Packet Switched Network (PSN) is
event and have demonstrated their ability to another key requirement for the migration of Mobile
transport E1 TDM data over their microwave links, Backhaul to packet switched networks. During the
namely Alcatel-Lucent 9500 MPR, Ceragon FibeAir event two implementations of ATM transport over
IP-10, and NEC PASOLINK NEO. The Alcatel-Lucent MPLS as specified in RFC 4714 were tested and
9500 MPR microwave solution performed at the demonstrated.
same time circuit emulation service. Successful interoperability was tested between
RAD Data Communications demonstrated IETF circuit Nokia Siemens Networks Flexi WCDMA BTS and
emulation (SAToP, RFC 4553) over MPLS which is RAD ACE-3200. The two devices encapsulated ATM
very similar to the MEF8 specification. In one case data into a statically configured MPLS pseudowire
the Circuit Emulation Service was demonstrated and sent the resulting MPLS packets over an IP tunnel
between RAD ACE-3200 and RAD ACE-3205 in across the PSN. The ATM pseudowire was used to
MPLS metro network. The Ceragon FibeAir IP-10 transport an ATM circuit configured between Nokia
microwave solution was connected both between Siemens Networks RACEL, a Radio Network
the RAD ACE-3200 and E1 source, and also Controller (RNC) and mobile core network emulator,
between the RAD ACE-3200 and an MPLS metro and Nokia Siemens Networks Flexi WCDMA BTS.
edge device. Ceragon Networks and RAD Data In addition, RAD Data Communications demon-
Communications demonstrated a hybrid mobile strated a statically configured ATM pseudowire
backhaul network operation, effectively combining between the RAD ACE-3205 and RAD ACE-3400
native TDM transport and Ethernet encapsulated devices. This pseudowire was tunneled over the
CES. In another test case SAToP CES was demon- PBB-TE network.
strated between RAD ACE-3400 and RAD
ACE-3205 in PBB-TE network.
12. Mobile Backhaul
Physical Topology Diagram Harris Stratex
Alcatel-Lucent
9500 MPR Spirent
Eclipse TestCenter
Rohde & Schwarz SITLine ETH
RAD ADVA FSP
RICi-155GE 150CC-825
RAD
RICi-155GE
Alcatel-Lucent
7705 SAR
Alcatel-Lucent
SIAE MICROELETTRONICA ALS 1850 TSS-40
InfoVista
VistaInsight for Networks
Ericsson
NEC PASOLINK NEO Marconi OMS 2400
IXIA XM2
IxNetwork T-MPLS Metro Alcatel-Lucent
Nokia Siemens Networks 1850 TSS-40
FlexiHybrid Huawei
NE5000E
Cluster System
Alcatel-Lucent
Ericsson Ericsson 7750 SR7
Gaming Client Alcatel-Lucent
9500 MPR Marconi OMS 2400 Marconi OMS 2400
Huawei
Cambridge NE40E-4
Telco Systems IXIA XM2
T-Marc-254 VectaStar
IxNetwork
Juniper 12
M10i
Huawei
Video Client CX600-4
Spirent
Telco Systems TestCenter
T5C-24G
RAD Juniper
RICi-155GE Juniper MX480
Cisco MX480
ME3400-12CS
Symmetricom TimeProvider Harris Stratex Redback
5000 PTP Grand Master Eclipse SmartEdge 400
Telco Systems Cisco 7604
T-Metro 200
ADVA FSP
MPLS
MTU-s
Ciena
150CC-825
Spirent LE-311v
Ceragon SIAE
TestCenter
MTU-s FibeAir IP-10 MICROELETTRONICA ALS
Telco Systems Ceragon
T5C-24F FibeAir IP-10
Juniper
MX240 NEC
Telco Systems CX2600
T-Marc-250 Ceragon MTU-s
RAD
FibeAir IP-10
ETX-202A
+ MiRICi E1/T1
RAD
ACE-3205 RAD
RAD RAD ASMi-54
ETX-202A ACE-3200 RAD
Application Demonstrations + MiRICi E1/T1 RAD LA-210
ETX-202A
Gaming Clients
running on E-LAN service
Video Equipment
running on E-Tree service Symmetricom
TimeCesium 4000
13. Device Types
Gaming Client
Metro/Core Network Device
Actelis IXIA XM2 Aggregation Device RAN BS
ML658 IxNetwork InfoVista
VistaInsight for Networks RAN NC Tester
RAD Video Client Access Device
IPMUX-216/24 Harris Stratex P Router
Eclipse Radio Transmission Device Cluster System
Rohde & Schwarz
SITLine ETH Alcatel-Lucent
Actelis 95000 MPR
ML658 Telco Systems RAD
T-Marc 380 ACE-3400
RAD RAD
ETX-202A RICi-16
Tejas
Tejas TJ2030 RAD
TJ2030 OP-1551
NEC
Nortel
PASOLINK NEO TE
Tejas MERS 8600
RAD
Tejas TJ2030 Egate-100
TJ2030 Spirent
TestCenter
PBB-TE Metro ADVA FSP
Ciena LE-3300 NEC PASOLINK NEO RAD 150CC-825
ACE-3205
ECI SR9705
Nortel
MERS 8600 Ciena LE-311v
SIAE MICROELETTRONICA ALFO
Tellabs
8830 Nortel Telco Systems
IXIA XM2
IxNetwork MERS 8600 T5C-XG Telco Systems
Spirent T-Marc-254
Foundry TestCenter
NetIron XMR 8000
Alcatel-Lucent IXIA XM2
7450 ESS-6 IxNetwork Cambridge VectaStar
Cisco
7606 Ceragon ADVA FSP
FibeAir IP-MAX2 150CC-825
Telco Systems
T-Metro 200 Gaming Client
ECI SR9705 RAD
Nokia Siemens Networks MTU-s Connection Types
hiD 6650 ETX-202A
Gigabit Ethernet
Cisco RAD
Fast Ethernet
Huawei ME4500 ETX-202A
Metro CX600-4 TDMoNxE1/STM-1
Nokia Siemens Networks
Tellabs Flexi WCDMA BTS ATMoNxE1/STM-1
Cisco Telco Systems
8830 Catalyst 3750-ME T-Marc-340 SHDSL
Cisco RAD G.SHDSL.bis
MTU-s ACE-3200
ME-3400-2CS Wireless
Telco Systems Video Source 10 Gbit G709
T-Metro 200 STM-16
Alcatel-Lucent NEC PASOLINK NEO
7705 SAR MTU-s 10 Gbit Eth
10 MHz Clock
Nokia Siemens Networks
NEC RACEL
Rohde & Schwarz PASOLINK NEO TE
IXIA XM2 SITLine ETH NEC Network Areas
IxNetwork CX2600 Aggregation network
RAD
IPMUX-216/24 Alcatel-Lucent
Metro network
Calnex 5650 CPAM Access network
Paragon Sync Core network
UNI
E-NNI
14. Carrier Ethernet World Congress 2008 Multi-Vendor Interoperability Test
Clock Synchronization Introduction Precision Time Protocol (PTP); Nokia Siemens
Networks’ Flexi WCDMA BTS received the PTP
Across all mobile network technologies, clock packets as a slave clock over an E-Line across the
synchronization is a topic of interest and major backhaul network. Calnex Solutions connected their
concern today. Base stations within a mobile Paragon Sync in between the master and slave
operator’s domain require a common clock for three clock, witnessing the protocol exchange and inten-
reasons: tionally dropping packets.
• General operation and frequency stability. Base
stations need to keep their transmit frequencies
and time slots very stable to avoid interferences.
• Base station hand over. Voice calls shall not
drop when the cell phone is moving from one
cell’s coverage area to the next. Clock
frequencies of the two base stations need to be
synchronous to ensure that the phone can
continue sending within its pre-assigned mobile
network slots nearly uninterruptedly.
• Common frequency transmission. In some mobile
technologies, adjacent base stations transmit
Figure 7: 1588v2 Synchronization
using identical frequencies, leading to a large
measurement
common frequency coverage area and allowing
the communication of end systems with multiple
base stations at the same time (MIMO). This The master and slave clocks interoperated success-
network service requires phase synchronization fully using the subset of IEEE 1588v2 Precision Time
of base station clocks to ensure that their signals Protocol required for frequency synchronization —
do not extinguish each other. the unidirectional SYNC messages. Via an E1 output
The easiest way of providing a common clock is to of the base station and using a Symmetricom
use GPS. However, mobile operators do not always software, we examined the frequency accuracy of
prefer this solution due to technical or political the base station. Figure 7 shows the first 85 minutes
reasons. Sometimes base stations do not have of a clock deviation measurement of the Nokia
visibility of the sky (pico cells in buildings, tunnels) or Siemens Networks Flexi WCDMA BTS synchronized
the GPS receiver installation would be too via IEEE 1588v2 to the Symmetricom TimeProvider
14 cumbersome (femtocells at home). 5000 PTP Grand Master clock. We started the
measurement after the first five minutes of device
Network Type operation — the amount of time the devices require
Synchronization
Synchronization
for the initial clock synchronization. As shown in the
diagram there is a peak of frequency deviation in
Frequency
the first two minutes of the measurement, and
Phase
another peak at around 20–30 minutes. In any case,
the deviation never exceeded 3.6 ppb (parts-per-
billion). The deviation was most often measured at
GSM X around 0.6 ppb. The measured results demonstrates
the accurate synchronization of the Flexi WCDMA
3G (UMTS, CDMA2000) X
BTS, and the test goal to achieve a frequency
4G (LTE including MBMS) X X deviation of below 16 ppb was fulfilled. Although
we did not conduct long term measurements as
4G (Mobile WiMAX) X X required in ITU standards due to a lack of time at the
hot staging; the same test will be conducted live at
CEWC for visitors to witness the clock accuracy
Vendors offer a number of network-based clock maintained throughout the conference.
synchronization mechanisms. They can be selected
For the same reasons, PTP impairments generated by
depending on the precision requirements. For
the Calnex Paragon Sync did not show visible
mobile backhaul, the base station frequency needs
effects. The base station has an internal temperature-
to be accurate to below 50 ppb (ITU-T G.812). The
controlled quartz as a fallback clock when the
network clock needs to be three times more accurate
incoming IEEE 1588v2 signal is lost. This clock is
to reach this goal — 16 ppb (ITU-T G.8261 draft).
accurate enough for a couple of days of operation;
we lacked the time to wait for it to degrade.
Clock Synchronization Test Results Adaptive Clocking. Another solution for some
IEEE1588v2. For the first time in a public Mobile frequency synchronization scenarios is adaptive
Backhaul test, we verified interoperability of IEEE clocking. In this solution, the clock is regenerated
1588v2 based clock synchronization implementa- from the frequency of bits arriving on an emulated
tions at this event. Symmetricom provided a TimePro- TDM circuit. Assuming that the transmitter sends
vider 5000 PTP Grand Master implementing the packets at a known rate and precise intervals, the
15. Ethernet OAM
adaptive mode is usually accomplished on slaves by First Mile (EFM), and Connectivity Fault
either measuring packets inter-arrival time or Management (CFM), standardized by the IEEE
monitoring a buffer fill level (some adaptive clock under 802.3ah and 802.1ag respectively, Y.1731,
recovery mechanisms may also use timestamps). standardized by ITU-T, and E-LMI, specified by MEF.
Adaptive clock works only for constant bit rate Over the past years we have seen a significant
services. increase in support in this area – in the first event
As described in the "Circuit Emulation Service" four vendors tested their CFM implementations and
chapter, the Alcatel-Lucent 9500 MPR, NEC five vendors tested their EFM code. At our current
CX2600, NEC PASOLINK NEO TE, RAD event 12 vendors tested their EFM and CFM imple-
IPMUX-216/24, and RAD MiTOP-E1/T1 hosted by mentations.
the ETX-202A have successfully demonstrated and Our service provider panel placed a high value on
tested adaptive clocking. both Ethernet OAM test areas and with the inclusion
In addition we measured the value of the frequency of both EFM and CFM in the new MEF 20 “User
deviation as demonstrated between the RAD Network Interface (UNI) Type 2 Implementation
ACE-3200 and RAD ACE-3205 over a long Agreement” technical specifications a clear
measurement time. We connected the Symmetricom continuous need for testing has been established
TimeCesium 4000 Master Clock to the RAD
ACE-3200, and verified that the frequency accuracy
was better than 16 ppb.
Link OAM
ETHERNET OAM Link OAM is the name used by the Metro Ethernet
Forum (MEF) to refer to clause 57 of the IEEE 802.3
EANTC interoperability events have integrated standards where OAM is defined for Ethernet in the
Ethernet Operations, Administration and First Mile. The protocol monitors the health and
Management (OAM) testing since 2006. Several operations of the UNI’s physical layer. The MEF
different protocols fall under the category of Ethernet requires the usage of Link OAM between the UNI-N
OAM. In this event we have tested Ethernet in the and UNI-C starting from UNI type 2.2 and recom-
Cisco
RAD 7604 Cisco
RICi 16 ME-3400-12CS
RAD
ETX-202A Cisco
Catalyst 3750-ME 15
RAD
LA-210
RAD
MTU-s ADVA FSP
150CC-825
RICi
155GE
Actelis
Ciena ML658
LE-311v
MTU-s Spirent
TestCenter
Alcatel-Lucent
7450 ESS-6
IXIA XM2 Ceragon
IxNetwork FibeAir IP-10
Huawei
Telco Systems
CX600-4
T-Marc-340
Tellabs
8830 Telco Systems
Foundry T5C-24F
NetIron
XMR 8000 Telco Systems
Juniper MTU-s
MX480 Juniper Telco Systems T-Marc-250
MX240 T-Marc-380
Discovery followed by Access MTU Tester
Loopback Mode Device MTU-s Switch
Dying Gasp Messages Aggregation Metro/Core
Device Device
Figure 8: Ethernet Link OAM test results