Agenda u History u The of IP Backbones

Agenda u History u The of IP Backbones Emerging Two Layer Network u Network Platforms u Standards and Forums u GMPLS 1

The Emerging Networking Paradigm ØThe change in the fundamental character of backbone network traffic, as demonstrated by the current shift in the telecommunications industry from traditional voice-centric TDM/SONET circuit-switched networking paradigm to data-centric packet-optimized networking paradigm, is leading to revolutionary changes in the traditional concepts of how networks are constructed. ØThe primary reason is the nature of traffic crossing today's long-haul backbones. ØInternet Protocol (1 P) applications are the fastest growing segment of a service provider's network traffic. . 2

Typical IP Backbone (Late 1990’s) Core Router ATM Switch MUX SONET/SDH ADM SONET/SDH DCS SONET/SDH ADM MUX ATM Switch Core Router u SONET/SDH ADM Core Router Data piggybacked over traditional voice/TDM transport 3

4

Why So Many Layers? u Router u MUX Packet switching u Speed match router/ switch interfaces to transmission u Multiplexing and statistical network gain u SONET/SDH u Any-to-any connections u Restoration (several seconds) u Time division multiplexing (TDM) u ATM/Frame switches u Fault isolation u Hardware forwarding u Restoration (50 m. Seconds) u Traffic engineering u u u Restoration (sub-second) DWDM Raw bandwidth u Defer new construction u u Result u More vendor integration u Multiple NM Systems u Increased capital and operational costs 5

IP Backbone Evolution Core Router (IP/MPLS) u MUX FR/ATM Switch MUX SONET/SDH becomes redundant Core Router (IP/MPLS) v IP trunk requirements reach SDH aggregate levels v Next generation routers include high speed SONET/SDH interfaces DWDM (Maybe) 6 SONET/ SDH DWDM

IP Backbone Evolution Core Router (IP/MPLS) u Removal of ATM Layer v Next generation routers FR/ATM provide trunk speeds Switch v Multi-protocol Label Switching (MPLS) on routers provides traffic MUX engineering Core Router (IP/MPLS) SONET/ SDH SONET/SDH DWDM (Maybe) 7

Removing the ATM Layer Logical Topology u Why Remove ATM? v Two networks to manage - IP and ATM v Cell tax v Lack of high-speed SAR interfaces v High density of virtual circuits v IP routing protocol stress 8

Collapsing Into Two Layers IP Service (Routers) Optical Core Optical Transport (OXCs, WDMs, SONET ? ) 9

Collapsing Into Two Layers IP Service (Routers) Optical Core Optical Transport (OXCs, WDMs, SONET ? ) u IP router layer functions v v v v Service creation Multiplexing and statistical gain Any-to-any connections Traffic engineering Restoration (10 s ms) Subscriber to transport speed matching Delay bandwidth buffering and congestion control 10 Internet scalability

Collapsing Into Two Layers IP Service (Routers) Optical Core Optical Transport (OXCs, WDMs, SONET ? ) u Optical transport layer functions v TDM and standard framing format v Fault isolation and sectioning v Restoration (10’s ms) v Survivability v Cost efficient transport of massive bandwidth (DWDM) v Long haul transmission distances v Metro transmission distances ? ? 11

The Emerging Two-Layer Network Data Layer Routers IP Services Transport Layer OXC’s TDM’s WDM’s LH Transport Reduced cost u Transport layer visible to IP Services u Transport layer signaling is an open standard (RSVP & CR-LDP) u Reduced complexity u Network more scalable u Uniform admin & management of IP and transport layers u 12

SONET/SDH Benefits u Rapid and predictable restoration v 10 s of ms; depends on ring size v Simple to engineer Standard framing and multiplexing (Time Division Multiplexing [TDM]) u Maintainability u v Performance monitoring v Fault isolation and sectioning v Bandwidth management v Network management u Transparency v Voice, video or data traffic u Challenge v Remove complexity and keep benefits 13

SONET/SDH Benefits u Rapid and predictable restoration v 10 s of ms; depends on ring size v Simple to engineer Standard framing and multiplexing (Time Division Multiplexing [TDM]) u Maintainability u v Performance monitoring v Fault isolation and sectioning v Bandwidth management v Network management u Transparency Traffic Quickly Rerouted After Failure v Voice, video or data traffic u Challenge v Remove complexity and keep benefits 14

SONET/SDH Limitations u Traditional SONET/SDH-based networks v Engineered for voice, not data v Slow to provision u Planning complexity u Grooming complexity u Delivery measured in weeks v Expensive to scale u Space, power, one wavelength per chassis v Inflexible u Static not dynamic bandwidth u Granularity – why not 5. 5 Gbps ? v Little interoperability at “control plane” u Customers forced to buy from one vendor u Stifles “best-in-class” deployment u Packet layer – no visibility into optical layer 15

What is an IP Router? A Device Which Moves IP Datagrams Across an Internetwork From Source to Destination u Minimum qualifications ISO 7 Layer Model v Capable of switching IP datagrams: L 3 forwarding v Symmetric any-port-to-any-port switching speed v Delay-bandwidth buffering, plus congestion control v Internet scale IS-IS, OSPF, MPLS, BGP 4 7 - Application 6 - Presentation 5 - Session 4 - Transport u Today’s benchmark 3 - Network v Wire-rate forwarding on all ports 2 - Datalink v v 1 - Physical v v for 40 -byte packets Performance independent of load Support of Co. S queuing, shaping, and policing Traffic engineering Classification and filtering at wire rate 16

What is an IP Router? Routing Algorithm Goals u Optimal routes v Calculate and select the best routes – many methods ISO 7 Layer Model 7 - Application u v Functional efficiency with low routing protocol overhead 6 - Presentation 5 - Session u in a variable environment (hardware failure, high load, topology changes) u Rapid convergence v Slow route calculations cause loops 2 - Datalink 1 - Physical Robust and stable v Predictable and correct functionality 4 - Transport 3 - Network Simplicity and drops in service u Flexibility v Speed + accuracy to adapt to network changes (bandwidth, delays, queues, traffic levels, etc. ) 17

What is an IP Router? IP Service Creation u Any-to-any connectivity v Internet scale routing allows anyone ISO 7 Layer Model to connect to anyone (within or outside of own company) 7 - Application 6 - Presentation u v Processing granularity to differentiate HTML from FTP 5 - Session 4 - Transport u Multicast v Not possible with voice circuit switching technology v Internet radio, video on demand, push Web 3 - Network 2 - Datalink 1 - Physical Applications u Content sites v v v Directing Web traffic Complementing cache servers Security 18

Optical Cross-connects (OEO) SONET/SDH Digital Cross-connect (DXC) Also known as Digital Cross-connect Switch (DCS) DXC/DCS 19

Optical Cross-connects (OEO) SONET/SDH Digital Cross-connect (DXC) Also known as Digital Cross-connect Switch (DCS) ATM Electrical Switch Matrix STS-N DS-3 STS-N ATM STS-N DS-1 ATM DS-3 DS-1 STS-N STS-1 STS-N ATM DS-1 DS-3 20

All Optical Cross-connects (OOO) All Optical Cross-connect (OXC) Also known as Photonic Cross-connect (PXC) OXC/PXC 21

All Optical Cross-connects (OOO) l 2 l 4 All Optical Cross-connect (OXC) Also known as Photonic Cross-connect (PXC) l 1 l 3 Optical Switch Fabric l 3 l 4 l 1 l 2 22

What is an Optical Cross-connect? u Connects ISO 7 Layer Model 7 - Application 6 - Presentation 5 - Session one port (l) to another port u Add/Drop function with certain l u Delivers high bandwidth u Quick to provision bandwidth 4 - Transport 3 - Network l 1 Port 3 l 2 2 - Datalink 1 - Physical Port 2 Port 4 l 2 l 1 23

OXC/PXC Switching Mechanisms Micro-electrical Mechanical Systems u MEMs u Fibers v Used for many other applications Reflector From Lucent, Corning, Xros (Nortel), and others u Currently 8 x 8 OXC u 256 mirrors, long-term goal 1, 024 u Imaging Lenses v OXC v ADM uses seesaw MEMS u Electrical controls v Voltage applied to mirror; tilts on 2 MEMs tilting mirrors axis + or – 6 degrees u Switch times typically 10 to 25 ms 24

OXC/PXC Switching Mechanisms u Liquid Crystal Light Valves v From Spectra Liquid Crystal Cell Switch and Chorum ON technologies Output 1 Input v Switch speed sub-millisecond v Future switch speed in nanosecond v 1 x 2 port switch Polarizing Beam v 2 x 2 Add/Drop Splitter v Electrical controls Liquid Crystal Cell 25

OXC/PXC Switching Mechanisms u Liquid Crystal Light Valves v From Spectra Liquid Crystal Cell Switch and Chorum technologies Input v Switch speed sub-millisecond v Future switch speed in nanosecond v 1 x 2 port switch Polarizing Beam Output 2 v 2 x 2 Add/Drop Splitter v Electrical controls OFF Liquid Crystal Cell 26

OXC/PXC Switching Mechanisms Bubbles From Agilent 32 x 32 or dual 16 x 32 ports u Suitable for u u v Wavelength Interchange Cross-connect (WIXC) v Wavelength Selective Cross-connect (WSXC) v Optical Add/Drop Multiplexers (OADM) u Inkjet + Silica Planar Lightwave Circuitry v Electrical controls v Bubbles created by heating “index matching fluid” u Switch times under 10 ms 27

Developing an All Optical Packet Router u Needs v How do you read a photonic header? u The “pipeline” approach? v Switching and logic u Current technology not fast enough u Lithium Niobate devices have speed, but too much crosstalk u Photonic Bandgap Devices (optical equivalent to transistor) v Buffering/Memory u Optical buffers (fixed loop delays) exist, but are insufficient u Bi-stable lasers u Holographic memories u SEEDS (Self Electro-optic Effect Devices) 28

Operational Approaches: Overlay and Peer Models u Overlay model v Two independent control planes u IP/MPLS routing u Optical domain routing v Router is client of optical domain v Optical topology invisible to routers v Routing protocol stress – scaling issues v Does this look familiar? u Peer model v Single integrated control plane v Router and optical switches are peers v Optical topology is visible to routers v Similar to IP/MPLS model 29 ?

Operational Approaches: The Hybrid Model u Hybrid model v Combines peer & Overlay u Middle ground of 2 extremes u Benefits of both models v Multi admin domain support u Derived from overlay model v Multiple technologies within domain u Derived from peer model Peer UNI 30

Standards and Industry Forums u Optical Internetworking Forum (OIF) v Industry forum v Kick-off meeting May 1998 v Standard OIF UNI based on IETF work (CR-LDP/RSVP) u Internet Engineering Task Force (IETF) v Driving GMPLS standards development u Initial application was MPlambda. S v Peer model and Hybrid model v Extend MPLS traffic engineering to the optical control plane u u u Rapid provisioning Efficient restoration ITU-T v Study Group 13 v Study Group 15 31

IETF u GMPLS now Hosted by CCAMP WG v Common Control And Measurement Plane MPLS WG revised charter (without GMPLS) u Eleven main GMPLS building blocks u v Internet Drafts Current work includes extending existing control protocols (example, OSPF & ISIS) u New & future extensions considered u v BGP 4 u For cross AS, and Carrier of Carriers applications v LCAS u Link Capacity Adjustment Scheme protocol for SONET u SONET Virtual Concatenation (dynamic TDM circuit control) u Intent to submit work to ITU-T 32

ITU-T u Study Group 13 (SG 13) v Focus: Multi-protocol & IP-based networks & their inter-working u Study Group 15 (SG 15) v Focus: Optical & other transport networks v G. ASON – Automatically Switched Optical Network u Addresses networks u Ambition the control layer for intelligent optical to reference IETF standards 33

OIF Optical UNI Signaling OIF-UNI UNI IETF-GMPLS UNI Optical Transmission UNI Network Uses procedures and messages defined for MPLS traffic engineering and GMPLS u Features u v Runs in UNI-only mode (overlay model) v Optical path creation, modification, and deletion v Optical path status inquiry and response u Allows one protocol to support two different applications v OIF UNI: client bandwidth requests (hide optical topology) v GMPLS: service provider provisioning (expose optical topology) 34

Traditional MPLS Applications Traffic Engineering Source Destination Layer 3 Routing VPNs CPE FT/VRF Traffic Engineered LSP PE PE P Site 1 FT/VRF CPE Site 3 FT/VRF CPE P Site 2 P CPE FT/VRF Site 3 FT/VRF PE P 35 PE FT/VRS FT/VRF Site 1

Generalized MPLS (GMPLS) Traditional MPLS supports packet & cell switching u Extends MPLS to support multiple switching types u v TDM switching (SDH/SONET) v Wavelength switching (Lambda) v Physical port switching (Fiber) Peer model u Uses existing and evolving technology u Facilitates parallel evolution in the IP and optical transmission domains u Enhances service provider revenues u v New service creation v Faster provisioning v Operational efficiencies 36

GMPLS Mechanisms IGP extensions u Forwarding adjacency u LSP hierarchy u Constraint-based routing u Signaling extensions u Link Management Protocol (LMP) u Link bundling u 37

IGP Extensions OSPF and IS-IS extensions u Flood topology information among IP routers and OXCs u New link types u v Normal link (packet) v Non-packet link (TDM, l, or fiber) v Forwarding adjacency (FA-LSP) 38

IGP Extensions OSPF and IS-IS extensions u Flood topology information among IP routers and OXCs u New link types u v Normal link (packet) v Non-packet link (TDM, l, or fiber) v Forwarding adjacency (FA-LSP) 39

IGP Extensions New Link Type sub-TLVs u Link protection 1: 1 Protection v Protection capability v Attributes u None, 1+1, 1: N, or ring u Priority for a working channel Working Protection 1: 3 Protection Working Protection 40

IGP Extensions New Link Type sub-TLVs u Link descriptor 1: 1 Protection v Characteristics of the link v Selected attributes Working u Link type v SONET, SDH, clear, Gig E, 10 Gig E Protection u Minimum reservable bandwidth u Maximum reservable bandwidth 1: 3 Protection v Attributes change over time Working v Provides a new constraint for LSP calculation u Shared Risk Link Group (SRLG) v List of the link’s SRLGs v Does not change over time 41 Protection

Forwarding Adjacency Ingress Node (Low Order LSP) Egress Node (Low Order LSP) SONET/SDH ADM ATM Switch u. A FA-LSP Ingress Node (High Order LSP) SONET/SDH ADM Egress Node (High Order LSP) node can advertise an LSP into the IGP v Establishes LSP using RSVP/CR-LDP signaling v IGP floods FA-LSP v Link state database maintains conventional links and FA- LSPs A second node wanting to create an LSP can use an FA-LSP as a”link” in the path for a new, lower order LSP u The second node uses RSVP/CR-LDP to establish label bindings for the lower order LSP u 42 ATM Switch

Forwarding Adjacency Ingress Node (Low Order LSP) ATM Switch SONET/SDH ADM FA-LSP Ingress Node (High Order LSP) SONET/SDH ADM Egress Node (Low Order LSP) Egress Node (High Order LSP) u IGP attributes describing a forwarding adjacency v. Local (ingress) and remote (egress) interface IP addresses v. Traffic engineering metric v. Maximum reservable bandwidth v. Unreserved bandwidth v. Resource class/color (administrative groups) v. Link multiplexing capability (packet, TDM, l , or fiber) v. Path information (similar to an ERO) 43 ATM Switch

LSP Hierarchy PSC Cloud TDM Cloud LSC Cloud Fiber 1 Fiber n LSC Cloud TDM Cloud l LSPs Time-slot LSPs PSC Cloud Bundle FA-PSC FA-TDM Explicit Label LSPs Time-slot LSPs FA-LSC l LSPs Fiber LSPs (Multiplex Low-order LSPs) Explicit Label LSPs (Demultiplex Low-order LSPs) Nesting LSPs enhances system scalability u LSPs always start and terminate on similar interface types u LSP interface hierarchy u v v Fiber Switch Capable (FSC) Lambda Switch Capable (LSC) TDM Capable Packet Switch Capable (PSC) 44 Highest Lowest

Constraint-based Routing Extended IGP Routing Table Traffic Engineering Database (TED) Constrained Shortest Path First (CSPF) u Reduces the level of manual configuration u Input to CSPF Explicit Route v Path performance constraints v Resource availability v Topology information RSVP Signaling (including FA-LSPs) u Output v Explicit route for GMPLS signaling 45 User Constraints

GMPLS Signaling Extensions Label Related Formats (“Generalized Labels”) u Generalized label request v. Link protection type (none, 1+1, 1: N, or ring) v. LSP encoding type (packets, SONET, SDH, clear, DS-0, DS-1, …) u Generalized label object v. Packet (explicit in-band labels) v. Time slots (TDM) v. Wavelengths (lambdas) v. Space Division Multiplexing (fiber) u Suggested label v. Label can be suggested by the upstream node v. Speeds LSP setup times u Label set v. Restrict range of labels selected by downstream nodes v. Required in operational networks 46

GMPLS Signaling Extensions PATH SONET/SDH ADM u RESV SONET/SDH ADM Bi-directional LSPs v Resource contention experienced by reciprocal LSP using separate signaling sessions v Simplifying failure restoration in the non-PSC case v Lower setup latency u RSVP notification messages v Notify message informs non-adjacent nodes of LSP events v Notify-ACK message supports reliable delivery u Egress control v Terminate LSP at a specific output interface of egress LSR 47

Link Management Protocol LMP LMP Control Channel Bearer Channel u The link between two nodes consists of v. An in-band or out-of-band control channel v. One or more bearer channels u Link Management Protocol (LMP) v. Automates link provisioning and fault isolation v. Assumes the bi-directional control channel is always available u Control channel is used to exchange v. Link provisioning and fault isolation messages (LMP) v. Path management and label distribution messages (RSVP or CR-LDP) v. Topology information messages (OSPF or IS-IS) 48

Link Management Protocol Services Provided by LMP u Control channel management v Lightweight keep-alive mechanism (Hello protocol) v Reacts to control channel failures u Verify physical connectivity of bearer channels v Ping test messages sent across each bearer channel u Contains sender’s label [(fiber, λ) pair] object for channel v Eliminates human cabling errors u Link property correlation v Maintains a list of local label to remote label mappings v Maintains list of protection labels for each channel u Fault isolation v “Loss of light” is detected at the physical (optical) layer v Operates across both opaque (DXC) and transparent (PXC) network nodes 49

Link Bundling Bundled Link 1 Bundled Link 2 u Multiple parallel links between nodes can be advertised as a single link into the IGP v. Enhances IGP and traffic engineering scalability u Component links must have the same v. Link type v. Traffic engineering metric v. Set of resource classes v. Link multiplex capability (packet, TDM, λ, port) bandwidth request) (bandwidth of a component link) u Link granularity can be as small as a λ u (Max 50

GMPLS Benefits u Open standards allow selection of best-in-class equipment u Routers have visibility into the transmission network topology v Eliminates N 2 meshes of links scaling issue v Reduces routing protocol stress v Optical paths span an intermix of routers and OXCs to deliver provisioning-on-demand networking u Leverages operational experience with MPLS-TE u No need to reinvent a new class of control protocols u Promotes parallel evolution of UNI and NNI standards u Enables rapid development & deployment of new OXCs 51

GMPLS: Modern Thinking for Modern Times Aligns with the way that the next generation network needs to be built and managed u 20 th Century – Transmission network was dominant u v Voice ran over the transmission network v ATM/Frame Relay delivered private data services v Internet was just one among many services v Transmission network created subscriber services u 21 st Century – Internet is dominant v Routers create the services that matter ($) v Network must be optimized for IP/Internet v OC-48/OC-192 make routers the largest consumers of bandwidth v New architecture is driven by routers subsuming functions previously performed by the transmission network u The transmission network must evolve in a way that is most beneficial to the creation of Internet services 52