[PDF]Unified MPLS - Rackcdn.com249459ec2cf8839ca4b8-2690f879103214107f979ba5105d745b.r47.cf2.rackcdn.co...
24 downloads
301 Views
13MB Size
EVOLVED PROGRAMMABLE NETWORK FOR MOBILE BACKHAUL SERVICES ALEXANDER PREUSCHE CSE SP ARCHITECTURES
03 NOVEMBER 2014
Agenda • EvoluBon and Trends in Mobile Networks – Requirements with LTE/LTE-‐A & Small Cells
• • • •
Evolved Programmable Network & Unified MPLS EPN Product PorRolio Key InnovaBons with SDN & NfV Summary
EVOLUTION AND TRENDS IN MOBILE NETWORKS
The Future of Mobility – 2018 perspecBve By 2018, mobile data traffic per month will reach
15.9 Ebs
11-‐fold growth from 2013 1000-‐fold from 2005
By 2018, over
50% of Mobile Traffic will come from 4G
By 2018, there will be more
10 billion 50 billion connected than total mobile-‐ready devices things by 2020 -‐ IoE
By 2018, 2/3 of the world’s mobile data traffic will be
video
Streaming Audio account for 10%+ Source: Cisco Visual Networking Index 2014 http://www.cisco.com/c/en/us/solutions/service-provider/visual-networking-index-vni/index.html
Access Technology – Defined by Traffic Growth Macro
18x Growth Ubiquitous Coverage
Growth
1000 100
2G/3G/4G
Macro Capacity
10
Spectrum 1
1990
1995
2000
2005
2010
Source: Agilent
2015
Overall Capacity Not Keeping Pace with Data Demand
High Bandwidth
Consumer
Business
Community
Small Cells Increase ExisBng Capacity & Coverage
Comprehensive architecture combining licensed and unlicensed
Small Cell Backhaul Requirements Aspect
Small Cell Requirement compared to Macro Cells
LocaBon
InstallaBon at Street Level LocaBons
Coverage & ConnecBvity
Last mile coverage to small cells at street level as opposed to roofop
QoS Support
User Experience on Small Cell should be same as on Macro cells
Capacity Provisioning
Higher Spectral Efficiency; lower mean throughput than macro during busy Bme but similar peak
Availability & Resiliency
Lower availability is acceptable when enhancing macro capacity (overlapping coverage)
SynchronizaBon
Similar BH Sync requirements as Macro; Phase required for TDD-‐LTE
Security
IPSec mandatory when BH over untrusted network
OAM
Consolidated management important for high scale Small Cells
Small Cell Backhaul Architecture Fiber connected Drop & ConPnue
NLoS/ LOS
ASR901S ASR901S
ASR901S
ASR901S
Macro Site Router ASR901
ASR901S
NLoS
ASR901S
P2MP NLoS
fiber
ASR901S
Macro Site Router ASR901 ASR901S
mmW connected Drop & ConPnue
NLoS ASR901S
ASR901S
IP/MPLS Transport
Pre-‐aggregaPon IP/MPLS ASR903 Transport
mmW ASR901S
Macro Site Router ASR901
Fixed (DSL/GPON/DOCSIS)
Aggregation Node ASR 903/901
IP/MPLS Transport AggregaPon ASR9000
Mobile Core Node
Various Backhaul Access Technologies Transport type E1/T1
Capacity
Latency
n x 2Mbps
<10mS
Direct Fiber
100Mbps-‐Gbps
<1mS
FTTx (GPON)
2.5Gbps/1.5Gbps
1-‐7mS
285Mbps/105Mbps (75Mbps symmetrical)
<20mS
xDSL (VDSL2)
Rate reach dependent >100Mbps/20Mbps for 2Kf (bonded VDSL2)
5-‐15mS
uWave
>1Gbps
<1mS
Sub-‐6 GHz
<500Mbps
DOCSIS3.0
5mS (1-‐way)
Air Range
2-‐4km <500m
LTE Any-‐to-‐Any ConnecBvity & X2 Interface Requirements • In R8/9 – Today, X2 is mainly used for:
- Mobility management/Handover - Load management/Error reporBng
X2
• LTE Advanced (R10/11) introduces CoMP and ICIC: - - - -
CoMP X2
X2
Coordinated MulBpoint (CoMP) Improve cell-‐edge and average sector data rates via dynamic coordinaBon between mulBple network nodes Centralized Joint Transmission – Single UE is served by mulBple eNBs UE reports Channel State InformaBon to eNBs – exchanged via X2
Requires direct, low latency x2 communicaPon! • X2 “transport" latency requirement:
R8: 20-‐80ms
SON & R11 CoMP: 1-‐5ms
- CoMP efficiency decreases already with “any” delay, i.e. 5ms delay =20% efficiency loss
LTE-‐A = Improved throughput, lower interference and latency Provides a compeBBve advantage to the SP allowing for new services
SynchronizaBon Requirements ApplicaPon
Phase/Bme accuracy is necessary
Frequency
Phase
UMTS/LTE FDD ResidenBal
NA / 250 ppb
NA
UMTS Metro Small Cell
NA / 100 ppb
NA
GSM/UMTS/W-‐CDMA
NA
CDMA2000
+/-‐ 3-‐10 uS
TD-‐SCDMA
+/-‐ 1.5 uS
LTE-‐FDD LTE-‐TDD LTE-‐A MBSFN LTE-‐A CoMP (Network MIMO) eICIC (HetNet CoordinaBon)
16ppb / 50 ppb
NA +/-‐ 1.5 uS +/-‐ 1 uS +/-‐ 0.5-‐1.5 uS +/-‐ 5 uS
InnovaBon to Address Service Providers Challenges
BUSINESS AGILITY
SDN – Open and Programmable at all Layers Simplify / Reduce Complexity
Service Orchestration
OPERATIONAL SIMPLICITY SDN
¥$£€
MONETIZATION
NFV
Service Orchestration – Customized Delivery Automation / Accelerate Time to Revenue
NFV – Elastic Resource Capacity Reduce Total Costs Across all Services
Mobile Backhaul Challenges Traffic Growth High Scale, and Capacity for Any Access (Macro, Small Cell Wi-‐Fi)
Convergence ElasBc Carrier Grade Transport Architecture for any Access
MonePze & DifferenPate Time to Market Programmable Simplify Management & Lower TCO
EVOLVED PROGRAMMABLE NETWORK & UNIFIED MPLS
Entering a New Era in the SP Network EvoluBon ISDN Frame Relay
APPLICATIONS
IP Core
QAM
ATM
DS0, DS1, DS3 Muxs
SMDS PSTN
Edge
EVOLVED SERVICES PLATFORM
X.25 Access/ Agg IPv6
EVOLVED PROGRAMMABLE NETWORK
TDM Era §
§ §
IP NGN Era Managed Managed § CommodiBzaBon of IP services
TDM rigidity limits new services, forces architectural shif Voice Centric
plus high traffic growth limits
profitability Configurable Configurable
§ Network migraBon to mulB-‐ Defined by reliability Apps f f N Apps IIndependent ndependent ooservice Network etwork transport & a single
§ §
Take advantage of the shif from staBc connecBvity to virtualized Orchestrated with Self-‐service service creaBon
§
EPN becomes he N flexible network fabric linking data centres with SP App t& etwork InteracBon networks
protocol
Command Command LLine ine nterface § IInterface Data Centric §
Defined by convergence and
Proprietary Proprietary scale
Autonomic, with Ccentric ontrol & Visibility Move into an applicaBons service environment
§
ApplicaBon centric ProgrammaBc Interfaces Well-‐known
§
Defined as programmaBc and dynamic
Open & Pluggable
Open Network Strategy
Open SDN/NFV Innovations for an Evolved Programmable Network On-‐Demand
Automated
Always “ON”
Dynamic Scale
Fully Virtualized
Intelligent Convergece
ACCELERATE
Policy
Services Anywhere
Open and Programmable
Real-‐Time AnalyBcs
ApplicaBon
Ultra HD
Seamless
InteracBon
Experience VM
VM
CDN
M2M
Core
ApplicaBons
OPTIMIZE
Edge
Open APIs
VM Service OrchestraBon Apps
Service Catalog Access
CORE
MONETIZE
£ € ¥ $
Open APIs
VM / Storage Control
Evolved Services PlaRorm NCS NCS
Cloud
EDGE
Access
Mobility Evolved Programmable Network
What is the Cisco Evolved Programmable Network (EPN)? Network as the Fabric • Converged • Physical and Virtual
IoE
Secure and Resilient
• IPv6 • SoluBon scale with mulB-‐chassis and nPower
• nLight protecBon and restoraBon • Encrypted Transport • IPSec
Programmable and Virtualizable • Open APIs • VNFs
CDN
ApplicaBons
Open APIs
VM Service OrchestraBon Apps
Service Catalog
CORE
IPv6
Open APIs
VM / Storage Control
Evolved Services PlaRorm NCS NCS
EDGE
Access
Evolved Programmable Network
EVOLVED PROGRAMMABLE NETWORK
“Unified MPLS…classical MPLS with a few addiBons” Classical MPLS IGP/LDP Domain isolation
RFC 3107
BGP filtering
Flex Access
LFA R-LFA
BGP PIC
E2E OAM
L2/IGP/BGP/MPLSTP/LDP DoD
Unified MPLS
Architecture
Scalability
Security
Simplification
Multi-Service
U-MPLS
Unified MPLS Architecture
IGP/LDP Label BGP3107 Label Service Label
iBGP/eBGP
Access Node
EPC Gateway
Pre-Aggregation Node
Access Node
Access Network
Aggregation Network
IGP/LDP
Aggregation Node
Core ABR
IGP/LDP
L2
Massive Scale 100,000+ Nodes!
Core Network
Flexibility
Any Media Any Access Any Service
Centralised RR IGP/LDP
Simplicity
Autonimic Networking LFA/R-‐LFA PRIME nV
Programmability
Netconf/Yang BGP-‐LS/PCEP
MPLS
Sample E2E Unified MPLS Architecture
RouBng IsolaBon and Label Stack for LSP between Pre-‐Agg. Node Loopbacks AggregaPon Network
Access Network
Core ABR (Inline RR)
Agg. Node IGP/LDP Label
Push
BGP3107 Label
Push
Swap
Core ABR (Inline RR)
MPC Gateway
Core ABR (Inline RR) Pop
Push
Centralised RR Swap
Swap
Pop
Access Network Pre-‐Agg. Node
Agg. Node
L2
ISIS Level 1/OSPF x
ISIS Level 2/OSPF 0
ISIS Level 1/OSPF x
L2 Access Node
Core Network
Agg. Node
Pre-‐Agg. Node
AggregaPon Network
Core ABR (Inline RR) Swap
Access Node
Agg. Node Swap
Pop
Service Label
LDP LSP
LDP LSP
LDP LSP
BGP LSP
No IGP route is propagated from AggregaBon to the Core. IGP area has routes for that area only plus routes to core ABRs. Only the core ABR’s are propagated from L2 to L1 •
LDP labels are used to traverse each domain and reach core ABRs
•
BGP labels are used by Labeled BGP PEs & ABRs to reach Labeled BGP PEs in remote areas
•
Service (e.g. PW) labels are used by Label BGP PEs
High Availability with Unified MPLS Access Network
Aggregation Network
PAN Inline RR ç next-hop-self è CSG
Core Network
Aggregation Network
CN-ABR Inline RR ç next-hop-self è
CN-ABR Inline RR ç next-hop-self è
iBGP IPv4+label
Access Network PAN Inline RR ç next-hop-self è
iBGP IPv4+label
iBGP IPv4+label
CN-RR RR
FTTB iBGP IPv4+label
iBGP IPv4+label
CSG
CSG MTG AGN-SE
FTTB
Mobile Packet Core
AGN-SE
CSG
SGW/PGW
MME iBGP Hierarchical LSP!
LDP LSP !
LDP LSP !
BGP FRR Edge <100 msec
LDP LSP !
BGP FRR Core <100 msec
LDP LSP !
LDP LSP !
LFA FRR, Remote-‐LFA FRR < 50msec
21
Unified MPLS Architecture Summary
Simplified MPLS Transport with E2E OAM, performance Access Layer
Cell Site
AggregaPon node
Etherne t uW
Cell site Router
AggregaPon Layer
Pre-‐AggregaPon Layer
PGW SGW
DistribuPon
Core Layer
node management, provisioning with seamless resiliency
Core node
Ring Fibre
Flexible L2 & L3 transport virtualisaBon to support GSM, 3G & LTE,
Sample RouPng Architecture Access Node
Access Network Access Node
IGP/LDP
iBGP/eBGP
wholesale & retail opBons
Core ABR
AggregaPon Network
EPC Gateway
Core Network Core ABR
Scale for MPLS transport and opBmal rouBng through RFC
New levels of L2
Pre-‐AggregaBon Node
AggregaBon Node
AggregaBon Node IGP/LDP
3107 with BGP hierarchical LSPs
Centralised RR IGP/LDP
LTE S1 and X2 MPLS VPN Service Scale Control Export: RAN W RT, Common RT Import RAN W RT, MTG RT
Core Domain
Aggregation Domain
Export: RAN Y RT, Common RT Import RAN Y RT, MTG RT
Aggregation Domain
MME VRF
MTG
VRF VRF
VRF
VRF
LTE Transport MPLS VPNv4/v6
MTG VRF
SGW/PGW VRF
VRF
VRF
SGW/PGW VRF
• Unified MPLS transport with a
VRF
MTG
VRF
Export: RAN X RT, Common RT Import RAN X RT, MTG RT
VRF VRF
Export: MTG RT Import: MTG RT, Common RT
VRF
VRF
Export: RAN Z RT, Common RT Import RAN Z RT, MTG RT
common MPLS VPN for LTE S1 from all CSGs and X2 per LTE region
• Mobile Transport GWs import all RAN & MPC Route Targets, and export prefixes with MPC Route Target • CSGs (and Pre-Aggregation Node) in a RAN region import the MPC and neighboring RAN Route Targets:
Enables S1 control and user plane with any MPC locations in the core Enables X2 across CSGs in the RAN region
Inter-‐Access LTE X2 – Labeled BGP MTG BGP Community 1001:1001
MTG MTG
CN-RR RR
CN-ASBR Inline RR
CN-ASBR Inline RR AGN-ASBR Inline RR
AGN-ASBR Inline RR
Metro-1
AGN-RR
Access-2
VRF
X2 Unified MPLS Transport: Advertise loopbacks in iBGP labeledunicast with community 10:10, 10:102
VRF
LTE MPLS VPN Service: Export: RAN-2 RT, Common RT Import RAN-1 RT, RAN-2 RT, RAN-3 RT, MTG RT
S1 traffic
RR
Inter-access X2 traffic
Access-4 Access-3
X2
VRF VRF
X2 inter-access
X2 VRF
VRF
VRF
VRF
Unified MPLS Transport: Advertise loopbacks in iBGP labeledunicast with community 10:10, 10:104
VRF
X2 inter-access
Unified MPLS Transport: Advertise loopbacks in iBGP labeledunicast with community 10:10, 10:103 LTE MPLS VPN Service: Export: RAN-3 RT, Common RT Import RAN-2 RT, RAN-3 RT, RAN-4 RT, MTG RT
LTE MPLS VPN Service: Export: RAN-4 RT, Common RT Import RAN-3 RT, RAN-4 RT, RAN-5 RT, MTG RT
Simplified MPLS VPN Scale Control for LTE Aggregation Domain
Core Domain
Aggregation Domain
Export: RAN Y RT, Common RT Import RAN Y RT, MPC RT
MME VRF
MTG
VRF VRF
Export: AGGR W RT, Common RT Import AGGR W RT, MPC RT
VRF
VRF
LTE Transport MPLS VPNv4/v6
MTG VRF
SGW/PGW VRF
VRF
VRF
VRF
SGW/PGW VRF
VRF VRF
MTG
VRF
Export: MPC RT Import: MPC RT, Common RT
VRF
VRF
Export: RAN Z RT, Common RT Import RAN Z RT, MPC RT
• Mobile Transport GWs import all RAN & MPC Route Targets, and export prefixes with MPC Route Target • CSGs in a RAN region import MPC and neighboring RAN Route Targets (Low Scale CSGs) or AGGR wide RT (High Scale CSGs) – Enables S1 control and user plane with any MPC locaBons in core – Enables X2 across CSGs in RAN and AGG region
The Autonomic Networking Infrastructure Secured Discovery and ConfiguraBon
Consistent Reachability
Security
a
Network
•
SUDI /UDI validation
•
Domain Certificates
•
Autonomic Control Plane
Discovery
•
Channel Discovery
•
Service Discovery
•
Autonomic Control Plane
•
Indestructible, virtual out-ofband channel
Auto-‐IP
Minimize Maintenance Windows & Touch Points
LLDP based Auto-IP negotiation
1
L2 Networks are popular in Access Rings since node insertion does not require adjacent node configuration
2
L3 Networks are challenging in Access Rings since node insertion requires adjacent node configuration
3
Auto IP solves this problem for L3 Networks by automatically assigning the IP addresses to adjacent nodes
Easy node insertion and IP address assignment in L3 rings Fast Service Deployments
Autonomic Network
Secured Discovery and ConfiguraBon 1
Configuration Engine
2
Auto-discovery and Secure Configuration Channel
3
4
Device shipped from Cisco manufacturing to branch with no configuration Device auto-discovered by neighbors and establishes secure configuration channel Device receives Configuration Engine location and securely registers Device downloads configurations from Configuration Engine
Zero-touch access auto-configuration
Autonomic MPLS Access Networks TFTP hosting Configurations co TFTP the AN Node
PANs
AN Connected NMS LAN
nfiguration
Unified MPLS Transport Registrar Gateway
Auto-IP Access Node
AN Virtual Out of Band Communication Channel
AN extended over IPv6 GRE tunnel
Syslog
AAA
• PAN and Access Nodes initiate a Virtual Out of Band (VOBC) communication channel automatically
• VOB channel relies on IPv6 link local addresses and RPL routing across VRFs and VLANs • PAN Nodes extend VOBC network space over GRE tunnels to Registrar Gateway • connects to AAA, Syslog, TFTP Servers and NOC • Access Node triggers an automatic configuration download and installation from a TFTP server. • TFTP address learnt from service discovery. • Auto-IP used in all configurations for seamless new Access Node activation
Ethernet G.8032 Access with Microwave ACM • The Ethernet Access Network adapts intelligently to the
Microwave Capacity drops
Aggregation Node
• Microwave Adaptive Code Modulation changes due to fading
events are signaled through an Y.1731 VSM to the MPLS Access Node
Aggregation Node
Ethernet interface
3. Policy Logic that updates G.8032 topology and H-QOS
• The Ethernet Access Nodes can trigger G.8032 failover
below a certain capacity threshold
• In addition the Access Node can change the Hierarchical 2. Y.1731 VSM Signals the Microwave link speed 1. Microwave Fading
QOS policy on the interface with the microwave system allowing EF traffic to survive despite of the capacity drop.
Microwave ACM Extension for MulBpoint Access EEM Event for specific Path with associated H-QOS actions
• ACM Signaling: The BW-VSMs are sent with Link ID TLV, specific to the impacted link
• The EEM Programmable Logic: Microwave Fading
BW-VSM with specific Link ID
Can classify events for specific paths Can adjust accordingly the H-QOS policies in this example
EPN PRODUCT PORTFOLIO
Deliver Ultra-‐High EPN MulB-‐Service Scalability Convergence without Compromise
Video High Scale by SoluPon Architecture
• Common high scale control plane • OpBmized forwarding resources • Scalable EFP-‐based service terminaBon
Business
MulP-‐service Hardware Design
• Line rate in the access • Per service HW structures at PE • Scalable H-‐QoS • MulBcast ReplicaBon • HW MAC Learning
Cloud
Mobile Modular IOS-‐XR
• Scale as you grow • Distribute processes between RP and LC • Ultra-‐high MulB-‐ Dimension Scale with superior stability
• • • •
HW Accelerated Ultra-‐ High BFD and EOAM Performance & Scale 3.3ms BFD 3.3ms CCM Fast failure detecBon Per LC scale
3.3 ms Unified MPLS
Evolved Programmable Network Family
MeeBng the Needs of Today’s Challenges and Tomorrow’s OpportuniBes
Multi-Service Core Routing:
UCS
Data Centre Dense 10GE/40GE/100GE
Drives 100GE multi-service density with MC Scale
LAN/SAN Switching Fabric
Nexus
Virtualized Compute
CRS
NCS: Edge Portfolio: Optimized 10GE/100GE Ethernet Density for Scalable Business, Consumer , Mobile, Video
Fixed and Mobile Convergence
ASR Series
Flexible Network Fabric Converging Core, Edge, Optical, Access, and Data Centre
NCS Elastic Access Portfolio: Converged TDM/Ethernet Aggregation
Access
GPON for wholesale and Mobile & Cloud demarc
Monetize IoE Opportunity
WAN + DC Physical and Virtual
Global Visibility and Programmability
Cisco Access and Small Aggregation Portfolio Snapshot Optimized platforms for Mobile, Carrier Ethernet access and aggregation deployment Updated in CY14
New in CY14
Carrier Class Aggregation
ASR 903
Modular chassis
ME 4600 New in CY14
Redundant switch processors
ASR 902
Wide selection of interface types
ME 3800X
Line Rate Performance, Multi-Dimensional Service Scale Coming CY14
FE/GE/10GE Ethernet access
ME 1200E
TDM interface Temperature hardened Low power consumption
New in CY14
ASR901S
ME 2600X
ASR901
ME 3600X
ASR 920 ME 3600X-24CX
ME 3400E
KEY INNOVATIONS WITH SDN & NFV
The Journey to true SDN/NFV Service InnovaBon Always “On”
Seamless Experience
On Demand Services Anywhere Application Interaction
IPv6
Networks
Networks
Simplify • • • •
Convergence / Consolidation Network Function Virtualization Service Chaining Service Orchestration
Accelerate New Services • • • • •
Bandwidth on Demand Virtual Managed Services Security Services Premium Mobile Broadband Cloud DVR
Networks Business Applications Integration The network proactively adjusts to the application needs in real time
NfV (Network FuncBons VirtualizaBon) Network infrastructure/Service FuncPons run on Virtualized x86 compute plaiorms Cisco UCS
•
Key Enabler: Cloud – –
•
Benefits: – – –
•
Hypervisor & x86 compute hardware Network automaBon / orchestraBon Faster service provisioning/Agility Shorter innovaBon cycle CAPEX & OPEX Savings
SDN complementary, but not mandatory
dDOS VM
SBC VM
Firewall VM
NAT VM
CGN VM
DPI VM
IPS VM
Virus Scan
DHCP VM
DNS VM
PCRF VM
Portal VM
WLC VM
RaaS VM
SDN Ctrl.
VM
BNG VM
NMS VM
Caching
CDN VM
WAAS VM
VM
VM
CE Architecture EvoluBon: towards Cloud-‐Centric Rapid growth of the applicaBons and services in the cloud à Rapid deployment of the transport pipe between users and the services in the cloud Transport Goal: Simple & programmable, Cloud integrated, Guaranteed SLA ESP Cross-‐domain OrchestraBon SDN Controllers
CPE vCPE NID
Access
AggregaBon
Core
AggregaBon
Cloud Edge (distributed NFVs) NaBonal DC
EPN (physical and virtual)
Regional DC
Access
CPE vCPE NID
Comparison of Some Leading SoluBons REP, G.8032, STP 802.1q/.1ad/.1ah
Fully distributed control plane (Unified MPLS)
L2 Bridging
Network Protocols
Complex • • • •
Unified operaBon across domains Full service, any scale, any topology, open, strict SLA Simple service provisioning But relaBvely complex transport
What if we fix this?
• • • • • • • •
Simple per-‐domain operaBon Rich porRolio, cheap? Complex cross domain operaBon, not end-‐to-‐ end, mulBple touch points Doesn’t support non-‐Ethernet service Limited network topology and scale Doesn’t support ECMP and TE L2 flooding L2TP caveats
Simple • • •
Simple network layer but complex controller layer Not mature for large scale deployment Service SLA? slow response to the network failures
Segment RouBng -‐ MPLS EvoluBon towards SDN
• Forwarding state (segment) established by ISIS/OSPF • LDP and RSVP-‐TE are not required • No need to migrate to IPv6 LDP/RSVP!
• MPLS Dataplane is leveraged without any modificaBon • push, swap and pop: all what we need • ECMP, PHP, normal h/w behavior
A
D
9001 M
O
N 72
Nodal segment: Operator allocates a label from the SR registry to each node. For example Z is given label 65 [any packet with 65 takes the shortest path to Z]
C
B
9001 65 Packet to Z
Adjacency segment: Node automaBcally allocates a local label for each adjacency. For example Label 9001 allocated for adjacency C-‐O [9001 is popped and packet takes this link to O] Combining nodal and adjacency segments as labels stack: The state is no longer in the network, it’s in the packet!
9001
Z P
65 Packet to Z
65
65 Packet to Z
Packet to Z
www.segment-‐rouBng.net
Transport Baseline: Segment RouBng • • • •
Segment RouBng, IGP shortest path as baseline, SR traffic engineering opBonally Any node to any node transport: SR node label Service node redundancy: anycast SR label Link or node protecBon by topology independent fast reroute (TI-‐FRR)
2 1 3
6 101
4 5
Core
102
7
Service Nodes Anycast label 1001
DC IGP/SR Domain: single area or process No IGP and LDP interacBon, NO hierarchy BGP and LDP LSP 50msec auto TI-‐FRR
Vision of the EPN Transport (CE) Architecture EvoluBon • Autonomic Network (physical): secure, auto discovery, plug-‐n-‐play • Segment rouPng (transport): 50msec self-‐protected, Agile on-‐demand TE • SDN controller (service): service label (or NSH) with cloud integrated SDN Controller DC Service label: 100 SR label: 5001
VSOC
ESP-‐CE
NFV Netconf/Yang Core [100,5001]
1 Anycast SR label: 5001 Service label: 100
2 CE Distributed DC NFV Simple, Agile, Programmable Transport
Transport node
DC APIC
NFV
Services moving to the cloud
Service/gateway node
Summary: the EPN Transport Architecture EvoluBon For the Era of the SDN/NFV/Cloud Is there a middle ground? Balance
Distributed
CE
BGP + SDN IGP SR
Controller Netconf/yang
Segment RouPng
Minimal but “Sufficient” distributed control plane intelligence Centralized intelligence on the SDN controller
Centralized
Thank you!