Transcript Slide 1

The Verizon NGN - Challenges in Evolving to a
Converged Network
Prodip Sen
Director, Packet Network Architecture
Verizon Technology Organization
June 1, 2007
Outline
• The Verizon NGN
• Packet Network Convergence
• Challenges and the Future
© Verizon 2007 – All Rights Reserved
Slide 2
The Verizon NGN
© Verizon 2007 – All Rights Reserved
Slide 3
Network(s) of the Past
STP
Switch
IXC
RT
DSLAM
Switch
Tandem
ATM
Switch
PSTN
FR
Switch
ATM
Switch
ATM/FR
Network
Gig-E
Switch
DCS
Inter-Office
Transport
Network
Cost of maintaining, growing and operating multiple technologies and networks is untenable.
© Verizon 2007 – All Rights Reserved
Slide 4
Services Landscape
• Demand for traditional wire line POTS is declining.
• Customers are becoming more technically sophisticated
with multiple devices requiring simultaneous broadband
access.
• The service model is changing from telephony-centric to
data-centric – most new services and applications being
developed, are IP / web based.
• “Any-to-any” connectivity - need to provide IP services to
(enterprise) customers with multiple locations served via
different Layer 1 or 2 access mechanisms.
• Quality of Service (QoS), Service Level Agreements
(SLAs) are becoming increasingly important.
• Service flexibility is the key to success.
© Verizon 2007 – All Rights Reserved
Slide 5
Business Drivers for Convergence
• Business Drivers
– Strategic Growth Services in which Vz will Invest
• Ethernet, Internet Access, L3VPNs, L2VPNs, VoIP & Video
Delivery
– Layer 1 and 2 services (Frame, TDM, ATM) will continue to exist for the
foreseeable future in their native form. Additionally these will provide
access to IP services.
– Services such as VoIP require rapid restoration and differentiated QoS
– FTTP will markedly increase traffic volumes
• Strategy
– Design a core “multi-service” network to serve all customer segments
– Use the converged network for new services - old technology will be
migrated and retired within financial and regulatory constraints
– Convergence to two layers in the core: optical transport and packet
switching, on which all applications can served.
– Complementing this evolution in the core network is the deployment
of FTTP for broadband access.
© Verizon 2007 – All Rights Reserved
Slide 6
Verizon’s Target – IP over Glass
•
Connectivity - Optical transport is the key to next-generation, bandwidth-intensive
applications.
–
–
FTTx is replacing the copper plant over the next 10-15 years.
Expansion of Verizon’s installed fiber plant via DWDM.
•
–
•
Evolution from ring-based SONET transport using APS protection to mesh-based DWDM
transport.
Access – Support legacy and new forms of access migrating to Ethernet
–
–
•
~3.5M strand-miles -> 3.5B λ-miles (pre-MCI merger)
Frame, ATM, TDM need to be supported, but will be pushed to the edge and aggregated
into the packet network.
Ethernet is the new access allowing for network convergence and significant savings.
Network - A QoS Enabled IP/ MPLS Network Provides Service Convergence.
–
Multiple overlay networks can be supported on a single core infrastructure, significantly
reducing capital and operating expenses.
–
–
–
•
Many levels of logical groupings possible - Virtual Private LAN Services (VPLS) and IP Virtual Private
Networks (VPNs), Logical Routers..
Aggregate forwarding in the core allows for significant scalability over traditional
technologies.
Class-based queuing in conjunction with MPLS allows for QoS-differentiated service
offerings, and quick failure-recovery.
Applications – Based on an IMS and IPTV infrastructure overlaid on top of the
packet network.
© Verizon 2007 – All Rights Reserved
Slide 7
Core Architecture Target
Next Gen
Elements
Applications
Service Delivery Platform
Network Apps
IP
MPLS
IMS Core;
IPTV Core
Converged
Packet Switch
/ Router
PBB, VPWS
SONET, EPL
G.709 (OTN)
Packet Optical
Transport
Platform
Mesh and Ring DWDM
© Verizon 2007 – All Rights Reserved
Slide 8
FTTP -- MASS-MARKET BROADBAND ACCESS
Super Head End
(SHE)
Central Office
Third Wavelength Is Optional
EDFA
VoIP
Services
Internet
Customer Premises
IP/MPLS
Network
IP Video
Services
O
N
T
OLT
PSTN/SS7
Network
Optical
Splitter
Optical Coupler
(WDM)
EDFA – Erbium Doped Fiber Amplifier
OLT – Optical Line Terminal
ONT – Optical Network Terminal
Industry Moving Towards GPON Systems
Bandwidth
Downstream:
Upstream
BPON
622 Mbps
155 Mbps
GPON
1.2Gbps/2.4 Gbps
622Mbps/1.2 Gbps
© Verizon 2007 – All Rights Reserved
–
–
–
–
Doubles, Or Quadruples, Bandwidth
Enables Full IPTV Implementation
Uses Same Fiber Plant Design
Overlay Wavelength Decision (Vs. IPTV)
Slide 9
Optical Network Convergence Strategy
• Approach
–
–
–
–
Consolidate traffic types into a single network
Reduce total number of network elements
Reduce number of optoelectronic conversions between ingress and egress points
Eliminate unnecessary regeneration in the network
• Integrated Multi-Service And Multi-Functional Elements in
Metro
– Service integration by providing full support for TDM/SONET/SDH, IP,
Ethernet, ATM, and MPLS interfaces
– Integration of aggregation, adaptation, switching, routing, and transport in A
high-performance, cost-effective design
• Transparent Optical Core
– ROADM and WXC Platforms in Core Network
– Mesh Topology – Dedicated Protection
© Verizon 2007 – All Rights Reserved
Slide 10
The Core Optical Network
Signaling Communication Network
Broadband
Service Control
Point
GMPLS Signaling and Routing
Messages
CPC
Wavelength Cross Connect (WXC)
CPC
CPC
CPC
Re-Configurable Optical
ADM
(ROADM)
CPC
Optical Cross Connect
(OXC)
Nationwide 10 Gbps Per
Wavelength Network (40G Ready)
Supporting Mesh Topology
Collector Rings
© Verizon 2007 – All Rights Reserved
Slide 11
Packet Network Convergence Strategy
• Platform Convergence
–
–
–
–
Reduce number of routers and interfaces to decrease Capex/Opex
L2 access network backhauls traffic to converged edge/aggregation routers
Multiple services supported via a converged edge/aggregation router
New services enabled by deploying new cards rather than new platform
deployment
– Common platforms enable convergence of testing, operations and OSS
Development for different business units
• Network Convergence
– Eliminate service specific networks, but maintain diverse customer access with
unified access into the VZ packet network
– Converge the backbone network and maintain logical networks based on class of
service sets rather than individual service
– Maintain logical control and capacity separation between service sets (e.g., public,
private)
© Verizon 2007 – All Rights Reserved
Slide 12
Strategic Packet Network Architecture
Verizon Business
Verizon Telecom
In Footprint
Access
Ethernet
NG VoIP,
Enterprise
Services
NG VoIP,
Consumer
Services
VoD/
IPTV
GPON
Under
Study
NGOLT
Next Gen
Edge/
Aggregation
Router
GPON
DSLAM
ROADM
Net
ADSL
Out of Footprint
Access
Converged
Back-Bone
Router
Multi-Service
Edge
Router
NGEAR
CPA
BEAS
Metro
Private
Line
MSE
LEC
TDM
FR/ATM
CBBR
TDM
ATM
NG-GWR
LEC
Ethernet
Access
Ethernet
Switch
Ethernet
SES
Metro
Fiber
FR/ATM
Legend:
Ethernet
FR/ATM
TDM
Optical
CPS
L2SW
CPA
L2SW
WAN
Lambda
Other
VPNs
VZ Wireless
External
© Verizon 2007 – All Rights
ReservedNetworks
CPA
BEAS
Internet
Slide 13
Packet Network Convergence
© Verizon 2007 – All Rights Reserved
Slide 14
Target Packet Network Characteristics
•
Service Support
– Uses MPLS VPNs and PseudoWires for service domain and customer
differentiation.
– Implements QoS to provide differentiated treatment of traffic types.
– Stable platforms and network resiliency mechanisms to provide PSTN-”like”
availability.
•
MPLS Model
–
–
–
–
•
IS-IS is the IP topology construction technology.
MPLS is the transport technology.
LDP is the initial MPLS signaling technology, with RSVP-TE phased in.
BGP is the VPN membership discovery technology.
QoS Model
– DiffServ combined with MPLS traffic engineering is used to provide end-to-end
QoS across multiple domains.
– NEs at the boundaries of a domain perform traffic control functions (e.g., policing,
marking, MPLS COS mapping).
– Interior NEs perform bandwidth management functions (e.g., aggregate queuing,
WRED).
© Verizon 2007 – All Rights Reserved
Slide 15
Convergence Enablers -- Router Design
• Carrier class routers emerging at last !
– Higher capacity, high availability, multi-chassis, diversity of service cards
• Software Process Separation
– Multiple routing processes run on the same physical processor with
operating system limits placed on key parameters
• Logical Interface Allocation
– Each logical interface (e.g., DLCI, VLAN) can be owned by a separate
process
• Hardware Processor Separation
– Each routing process runs on a physically separate processor
• Implementing Resiliency
– Fast Failure Detection (e.g. BFD)
– Non-stop forwarding via graceful restart or hot routing/signaling
redundancy
• Forwarding Separation
– Class-based queuing, scheduling, policing and shaping
– MPLS bandwidth reservation
– VPN-based forwarding
© Verizon 2007 – All Rights Reserved
Slide 16
Convergence Enablers -- Hardware and Protocols
• High-performance Ethernet Forwarding Hardware
– High-speed, cost-effective interfaces
– QoS capable, policing, shaping per logical interface
– Supports link aggregation, protection switching, OAM
• Ethernet-capable Optical Equipment
– New generation of optical aggregation and switching elements have
Ethernet and MPLS processing capability
• Tunnel-based Traffic Engineering and Constrained Routing
– MPLS support currently available
– Ethernet-based tunneling may be an expected future standard
• Automatic Logical Circuit Provisioning, Routing, Restoration
– Recent Multi-Segment Pseudowire (MS-PW) signaling and routing
standard provides scalability and inter-provider interconnection
– MS-PW protection and diversity routing being standardized as well
© Verizon 2007 – All Rights Reserved
Slide 17
Logical Router Technology – Separation/Allocation
Traditional Router
Software Separation
Processor
Routing
Process
Physical
Interface
Distributed Processing Router
Multiple Routing Processes
Single
Processor
Single
Routing
Process
Multiple Processors
SubProcesses
Hardware Separation
Multiple Routing Processes
Logical
Interfaces
Software
Separation
Hardware
Separation
Multiple
Processors
Switch
Forwarding
Cards
Logical
Interface
Allocation
Forwarding
Card
Allocation
Some
Separation,
Least Cost
Some
Separation,
Least Cost
Good
Separation,
Higher Cost
Most
Separation,
Highest Cost
Logical Interface Allocation
Multiple Routing Processes
Forwarding Card Allocation
Multiple Routing Processes
Logical
Interfaces
© Verizon 2007 – All Rights Reserved
Slide 18
Ethernet and MPLS Aggregation in the Access Network
Verizon Business
Verizon Telecom
In Footprint
Access
NGEAR
GPON
NGOLT
Internet
Access
L3VPN
Out of Footprint
Access
MSE
Internet
Access
L3VPN
LEC
TDM
BEAS
Tunnel
Ethernet VLANs
L2SW
MS-PW
Ethernet
Ethernet
Switch
Tunnel
L2SW
L2SW
MS-PW
Ethernet VLANs
L2SW
On-net
Fiber
Open Third
Party
Interface
• Multi-Segment Pseudowire (MS-PW) switching provides
any-to-any, automatic, traffic engineered virtual
connections
• MPLS or Ethernet Tunnels provide scalability within a
domain
• L2 protocol interworking supports connections with
different protocols at the end points
MS-PW
L2VPN
Ethernet
Network
Legend
MS-PW Segment Endpoint
MS-PW Segment
External Networks
© Verizon 2007 – All Rights Reserved
Slide 19
FTTP Access Aggregation – Functional Convergence
•
•
Optical transport technologies with integrated Ethernet switching (e.g. OTP) provide OPEX &
CAPEX reduction for traffic aggregation in the access network
Functionally decompose the edge GateWay Router into the NGOLT, OTP and the Next Gen
Edge/Aggregation Router
Current Deployment
SONET
ADM
GWR
LCR
OLT
Target Architecture
OTP
NGEAR
NGOLT
© Verizon 2007 – All Rights Reserved
Slide 20
Example : Splitting BRAS Functions Current View
Routing Aggregation
Access
Subscriber
Management
(policy, DHCP,
..)
L4+
L2-L3
GWR
Core:
routing,
forwarding, MPLS,
queuing, QoS, etc.
Edge routing
forwarding, 2547,
vlan, Diffserv Sub
queuing, policing,
etc.
OLT
Aggregation
L1
PON
© Verizon 2007 – All Rights Reserved
Slide 21
Example : Splitting BRAS Functions Target View
Routing Aggregation
IP-MPLS Service Edge
Access
Subscriber
Management
(policy, DHCP,
..)
L4+
L2-L3
Edge routing
forwarding, 2547,
vlan, Diffserv Sub
Core: queuing, policing,
routing,
etc.
forwarding, MPLS,
queuing, QoS, etc.
Forwarding, Diffserv
Subscriber-queueing,
Policing,
IGMP, Multicast
forwarding, AntiSpoofing, ARP
OLT
Aggregation
L1
PON
NGEAR
ROADM
Net
© Verizon 2007 – All Rights Reserved
NGOLT
Slide 22
Challenges and the Future
© Verizon 2007 – All Rights Reserved
Slide 23 23
Where Are We?
• We are attempting to
– Merge separate networks.
– Introduce fundamentally new technology in several areas simultaneously
• While
– Technology and standards are evolving
– Legacy technology and network elements remain and have to be cared for
• We need to
– Shift in thinking from circuit-switching to packet-switching.
– Change our operations paradigm and processes
• New IP technologies are more like the Internet, less like the PSTN.
• New technologies and strategies are forcing convergence in networks and
network elements.
• Multiple groups may need to touch the same elements and networks.
• Need help in the management plane – new technology dies on the vine
if not operationally viable
© Verizon 2007 – All Rights Reserved
Slide 24
Inter-Provider / Inter-Network Challenges
• Interconnection requirements driven by
– services requiring inter-provider connectivity and end-end QoS guarantees
(e.g., VoIP, global IPVPN services)
– the regulatory regime
– connecting existing networks
• Specifying and achieving performance across domains
– Common definitions of performance metrics across boundaries.
– Apportioning performance when traffic crosses multiple carrier networks.
– Enforcing SLAs across provider networks.
• Achieving resiliency across providers / networks
– MPLS is still optimized for intra-domain applications
– Inter AS control plane is designed for stability and scale – not
performance.
– Inter AS TE, Fast-Reroute, Inter-carrier OCh restoration technology are
still in their infancy.
• Troubleshooting across the boundary
© Verizon 2007 – All Rights Reserved
Slide 25
Beyond MPLS
• High speed forwarding
– Faster, bigger routers - are IP address lookups no longer an issue?
• Separation of control and forwarding
– Separate control processors and forwarding engines – is the separation
then just a matter of better network element design?
• Advanced services (VPN etc) and service separation
– Better implementations of routing contexts and logical routers – can these
be enough for service separation?
– Improved resiliency and TE
– Will IP fast reroute and the lack of useful tools to handle MPLS TE
complexity, overtake the use of MPLS?
• Encapsulating services
– Will handling legacy native layer 1 and layer 2 services via encapsulation
be the only reason left to use MPLS?
© Verizon 2007 – All Rights Reserved
Slide 26
Convergence and Simplifying the Core - Are We There Yet?
• In the current model for handling convergence are our
networks any less complex?
– E.g. the ATM control plane exists between CPE and bet end
switches, IP control plane between “core” routers and interworking
between the two at the boundaries.
– Legacy Layer 1 and 2 switching is preserved, together with the
new MPLS switching
– Issues with QoS mappings, path visibility, points of failure
• Are we moving complexity from the edge back into core?
– Are the next generation elements more complex failure prone
devices?
• Should we be building true label switches to simplify the
core ?
© Verizon 2007 – All Rights Reserved
Slide 27