1. On Evolving to Dynamic
Ethernet Lightpaths across
Optical Private Lines
Presenter: Roderick Dottin / Orange
Date: October 6, 2015
2. interne Orange2
Drivers
Meeting the ever growing performance demands of value-added
services with smaller and smaller margins. How can this trend be
reversed?
Transparency – Legacy multi-vendor GFP encapsulated Gigabit
Ethernet over VC-4-16c SDH Links
End-to end Performance – Carrier OA&M at OTN-encapsulated
boundaries and MEF 2.0 Ethernet SOAM across Operator Domain
Scale – Explore Network Operator / Orange Interworking at OTUk
Interface in order to extend capacity/reach of Backbone Network
3. interne Orange3
Hubbing Network
EoSDH
8 x 1GbE
US Domain – Transparent Ethernet over SDH (EoSDH)
Transport
VCG-1 (3 x AU3s); VCG-2 (7 x AU4s); …
8 x 1GbE
VCG-1 (3 x AU3s); VCG-2 (7 x AU4s); …
100 bT
1 GbE
100 bT
1 GbE
• Port-based VCAT mapping at international
multi-vendor GFP encapsulation nodes
• Bundle VCG constituents of various sizes into
AU4-16c paths across broadleaf network
AU4-16c AU4-16c
These virtual paths can evolve into dynamic scalable lightpaths within the
core optical network
Bundling restricts max. tolerable differential VCG delay to less
108ms (before generating Ethernet framing errors)
Above approach still allows for dynamic path restoration and optical
bypass capability
Ability to scale OTN client interfaces to 100G/400G and beyond
4. interne Orange4
Day 1 Solution – GFP Encapsulated Gigabit Ethernet over SDH
Manasquan
Tuckerton
NY/XFT6
___ South route
___ North route
Miami/XFT2
TAT-14 Sud
FLAG Atlantic-1
AM-II
Core Router
US Domain
EoSDH
SDH Gateway
PHL/XFT1
LOS OSOS
NY/XFT9
n x STM-64
n x STM-16 OJUS
Gemini /
CBUS
Island Park, NY
SONET Gateway
• Downgrading TAT-14 backhaul to
leased capacity will free-up
ROADMs
• To scale STM-64 links beyond
Day1 capacity, redeploy ROADMs
within US backbone
Japan / US
6 x GbE
**
:
5. Orange presentation
Reference Model – Infrastructure as a Service
Next Gen Stack…
–
• ITS
a = SDH / Sonet
b = Lambdas
c = Ethernet
Transport
• IP services
• ATM - QoS
Intelligent Optical Layer
ATM
a
b
ValueAddedServices
Layering Services
• Next Gen Services
• Voice Services
Hubbing, Mobile
ETHERNET
SDH /
SONET
OTN
VOICE INTERNET
Extracted from Ref [1], Slide 7
6. Orange presentation
International Access Ethernet Private Line Rollout
Carrier Pop-to-CHLS
Service Provider
Pop-to-CHLS
Orange
Backhaul
LL Orange
Backhaul
Carrier Pop-to-CHLS
Service Provider
Pop-to-CHLS
Orange
Backhaul
LLOrange
Backhaul
Submarine
Cable
Submarine
Cable
Submarine
Cable
Submarine
Cable
Extracted from Ref [1], Slide 9
7. Orange presentation
Operator B DomainOperator A Domain
Carrier OA&M at Core and Access Network
Boundaries
LF
LF
LF
RF
RF
ITU-T G.709 / Y.1331
(802.3ba Adopted
Baseline)
MEF 30 (Fault Mgmt per
IEEE 802.3ag / Y.1731)
MAC Status
Defects / Link Fault
Indication
MAC Status Defects /
Remote Fault
Indication
OTN Backbone
RF
LF
E-NNIE-NNI
UNI UNI
Service ProviderSubscriber Subscriber
Subscriber Domain
Service Provider Domain
LL Operator A LL Operator B
E-NNIE-NNI E-NNI
X
E-NNI
Carrier 2.0 Ethernet SOAM support over OTN – Ref [2], [3]
8. interne Orange8
Handling Scale in US Domain – Leased Optical Private Network
Manasquan
Tuckerton
NY / XFT6
___ South route
___ North route
Miami / XFT2
TAT-14 Sud
FLAG Atlantic-1
AM-II
Core Router
US Domain
EoSDH
SDH Gateway
PHL/XFT1
LOS OSOS
NY / XFT9
n x STM-64
n x STM-16 OJUS
Gemini /
CBUS
Island Park, NY
SONET Gateway
Japan / US
6 x GbE
**
:
WAS (ASHVA)
BMM
BMM
BMM
OTUk
NNI
OTUk
NNI
OTUn
NNI
ODUk
• Network Operator / Orange
Interworking at OTUk NNI
• Deliver 2.5G, 10G Services
as ODUk payloads
For my presentation today, I'll like to elaborate on an initiative to deliver international Ethernet services within the US Domain.
We’ll then explore next steps needed to evolve from the initial static approach, towards Dynamic Ethernet Transport
A major issue facing Carriers and Service Providers is having to deploy value-added services with ever growing performance demands yet smaller and smaller margins. A trend among various speakers at 2015 NGON USA conference is that we are reaching the point where costs will exceed revenue; especially with the rollout of data services. The question that begs answering is "How can this trend be reversed ?"
This presentation seeks to answer this question by taking a bottoms-up approach to the infrastructure rollout needed:
Starting with the day1 solution: Transparent transport is a prime driver (being an environment with multi-vendor nodes in International locations) so we use standards-based GFP encapsulation at the remote nodes, then employ VCAT mapping onto SONET/SDH payloads; in this case, concatenated STM-16's links;
For end-to-end performance, we must also consider carrier-grade OA&M capability across Network Operator boundaries;
Considering the need to scale, will explore vendor proposal of utilizing OTUk links to extend capacity/reach of the backbone network.
[BTW during the presentation the order of the 2nd and 3rd points will be reversed...]
In our bottoms up approach, we shall deconstruct the Day1 Solution. Shown here is the underlying structure for transparent Ethernet transport across the US Domain.
Incoming data traffic at Ethernet ports are encapsulated using frame-mapped Generic Framing Procedure (GFP-F) per ITU-T G.7041/Y.1303. These frames are then mapped onto SONET/SDH payloads using port-based Virtual Concatenation (VCAT) – per ITU-T G.707 11.1. This approach establishes a 1-to-1 relationship between ingress/Egress Ethernet ports and Virtual Concatenation Group (VCG) of contiguous time slots (constituents).
So for Ethernet ports of different sizes, the resulting VCG constituents of varying granularities are organized into dedicated AU4-16c paths/virtual circuits across the US backbone (bundling enabled in the grooming eqpt). The above has already been validated by the transparency testing done thus far.
For the planned evolution, these virtual paths can evolve into dynamic scalable lightpaths within core network; but with some embedded constraints:
bundling will restrict VCG differential delay to less than 108ms (avoiding the generation of framing errors);
dynamic path restoration and optical bypass will be possible but performed at 2.5G sub-wavelengths to keep the virtual circuits intact;
aligning the rollout with the established roadmap will allow for OTN Client I/Fs to scale to 100G/400G and beyond (as standards become available in the market).
Understanding the underlying infrastructure, we take a look at the implementation of the Day 1 Solution, exploiting leased capacity from the submarine cable landing stations, which is backhauled to the SDH backbone nodes.
So Gigabit Ethernet traffic from French West Indies – namely, Martinique and Guadeloupe – can be transported across the Americas II and CBUS submarine cables to the US backbone, then groomed onto TAT-14 South and FLAG transatlantic cables towards Paris; traffic from St Martin will terminate in NY at 111 8th Ave.
Note that transparency testing has been completed with multi-vendor eqpt across transatlantic link from France to New York, with zero Frame Loss and transmission errors.
While the existing network capacity can be met with STM-64 links, projected growth would require the stacking of leased STM-64 backbone links.
In parallel, a planned downgrade of the TAT-14 backhaul to leased capacity would free-up existing ROADMs, which could be redeployed within the US backbone to handle scale.
From the current static model we now transition to the planned evolution of the network. The reference model shown can support both legacy as well as Next-Gen services. It is based on an intelligent optical layer upon which Value Added Services such as Carrier Ethernet 2.0 and OTN wavelength services can be scaled and dynamically routed within the Core network; delivering: Next Gen, Managed Transport as well as existing legacy services (IP for Bus. Applications; ATM for QoS; Voice services).
There is ongoing convergence activity within the intelligent optical layer between SDN + OTN + Dynamic ROADM technology that will provide flexible, any-any grooming and delivery of user-driven, on-demand services – these services can be turned-up, modified and torn-down in real time.
Note that this collaboration could be the foundation for the Uberization of the backbone network and one of the reasons why NGON can be a pivotal environment in fostering such discussion.
Targeting Ethernet-Transport•as•a•Service, a reference circuit is defined for rollout of an international Access Ethernet Private Line (Access EPL). The design layout, as shown, provides a dedicated connection between two international cities from a LL Service Provider POP or the Carrier POP in City A across the pond to a Cable Head Landing Station (CHLS) in City Z. The circuit can then include a backhaul segment to the Carrier POP (which typically houses the ROADM eqpt) and a Local Loop segment to connect to the LL Service Provider. From the Service Provider's perspective, these are the nailed-up segments of the circuit with longest lead-times.
Once installed, it is the LL Service Provider who can now provide MEF 2.0 defined Access EPL services to Subscribers via managed demarc; maybe achieving Customer Prem to Customer Prem managed Ethernet transport via a tunneling mechanism such as Provider Backbone Transport (perhaps PBB-TE or MPLS-TP).
Such mechanisms will support: end-to-end connection protection, OAM packet transmission. These are carrier grade properties that can be brought to the customer premises.
Extending OAM capability to a managed demarc at Customer Premises remains a challenge; MEF 2.0 service definitions may make this possible. Carrier Ethernet 2.0 Service OAM architecture results in nesting arrangement in which a MEG with a lower MEG Level cannot exceed the boundary of a MEG with a higher MEG Level, defined such that:
1) a MEP at a particular MEG Level transparently passes SOAM traffic at a higher MEG Level;
2) terminates SOAM traffic at its own MEG Level;
3) discards SOAM traffic at a lower MEG Level.
Combining this Layer 2 architecture with the interoperability agreement shown, a failure within the OTN core will generate a Link Failure which is propagated across the core network to Ethernet interface at Egress. And per the MEG nesting defined above, is passed along Maintenance Entities within the Operator Domain to furthest E-NNI, at which Link Fault Indication is declared.
Likewise a Remote Failure is sent upstream within the core network to the Ethernet interface at Ingress, then passed along Maintenance Entities within Operator Domain to furthest E-NNI, at which Remote Fault Indication is declared; setting up a signaling mechanism for fault located within the core network.
Having laid out the infrastructure model and defined the reference circuit we return to the Day 1 configuration. For considerations to scale the network and expand ROADMs across the US backbone, it is now feasible to deploy leased Optical Private Network.
Instead of leasing dark fiber pair to reach a new backbone node, we lease a point-to-point OPN link with OTUk NNI hand-off between the Network Operator and the Orange ROADM; OTU4 is the planned interface.
The ODU1, ODU2 payloads will deliver the 2.5G and 10G services as virtual circuits. This template will be replicated across US backbone in turning up other nodes.
This concludes the planned evolution of the US backbone and I hope that you found it of interest. One more comment is that the discussion was derived from the ODIN initiative internal to Orange, but relaunched within the US domain with a more generic approach.
