This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
NGN Carrier Ethernet System Architecture Overview The Context of Broadband Forum’s TR-101 & MEF NGN Carrier Ethernet System Architecture Details ‒ Building Blocks and Variants ‒ Service Delivery Models ‒ Network-based High Availability ‒ Scaling with Unified MPLS
Massive change in SP traffic make-up in next 5 years* SP revenue shifting from circuits to packet services** ‒ 5 yrs ~80% revenue derived from packet services
‒ Packet traffic increasing at 34% CAGR*** *ACG Research 2011, ** Cisco Research 2010, ***Cisco VNI 2011
Cisco Carrier Ethernet Transport Architecture Technical Innovations
5
OAM Subsystem
IPoDWDM OpPcal Network
AggregaPon Node
AggregaPon Network MPLS/IP
Carrier Ethernet Aggregation!Access! Edge!
AggregaPon Node
AggregaPon Node
Ethernet Node
STB
VoD
Content Network
TV SIP
PON Node
DSLNode
Core Nodes
VoD
Content Network
TV SIP
EMS NMS Portal
AAA Service and Performance Mgmt DHCP, DNS
Multiservice Core!
Core Network IP / MPLS DistribuPon
Node
Corporate Business
Corporate Business
ResidenPal
STB
ResidenPal
AggregaPon Node
DistribuPon Node
Mobile
2G/3G/4G Node RAN Access Network
MPLS/IP
Corporate Business
BSC/RNC
BSC/RNC
MPLS-based transport with MPLS-TP option Cisco ASR9000, ASR1000, ASR 903, ASR 901, ME3600X/ME3800X/ME3600CX Cisco Carrier Packet Transport , CPT50, -200, -600
Flexible Ethernet Edge (EVC infrastructure code)
Ubiquitous Ethernet UNI across different product lines and OSs
Flexible Options for Subscriber Awareness Distributed Edge, Centralized Edge, ISG for IPv4/IPv6
Intelligent transport of video
PIM Optimizations, MoFRR, TI-MoFRR, integrated video caching, Video Quality Monitoring
Two technologies for L2 transport over MPLS: ‒ Ethernet over MPLS (EoMPLS)
Used for L2 point-to-point link over MPLS cloud No MAC learning involved
‒ Virtual Private LAN Services (VPLS) Used for multipoint L2 connections Collection of pseudowires tied together by a Virtual Forwarding Interface (VFI) MAC addresses learned on VFI Traffic forwarding based on destination MAC addresses H-VPLS, an extension of VPLS
Ethernet VPN is a new technology for NGN L2VPN Services (not described in this breakout!)
MPLS in the aggregation network and core Targeted LDP session between PEs to exchange VC label Tunnel label is used to forward packet from PE to P to PE VC label is used to identify L2VPN circuit Attachment Circuit (AC) can be port-based or VLAN-based (or Ethernet Flow Point
MPLS Transport Profile ‒ T-MPLS requirements feeding into IETF MPLS-TP enhancements:
MPLS-TP differs technologically from T-MPLS. ITU stopped work on T-MPLS. ‒ Effort to address Pt-to-Pt ATM-like transport centric networks (like ATM PVCs) ‒ Focused on connection-oriented (CO-PS) services
Data plane—based on IETF MPLS, with restricted options ‒ No ECMP, no PHP, no LSP merging
Control plane—static and/or dynamic ‒ Static provisioning with NMS, with standardized common functions ‒ Dynamic control plane based on GMPLS or IP/MPLS
Key OAM enhancements ‒ GE-ACH—Generic Associated Channel to support FCAPS functions alongside transport MPLS LSP ‒ GAL—Generic-ACH Label as generic exception mechanism for LSP OAM
TR-101 Scope and Content Migration from ATM to Ethernet Broadband Aggregation
15
VLAN architecture Multicast considerations Use of a video-optimised Service Router (next to ‘traditional’ TR-59 type
BRAS) Resilience in the Ethernet Aggregation Network QoS in the Ethernet Aggregation Network Ethernet OAM Support for PPPoA and IPoA (aka interworking between XoA and XoE)
Note: TR-101 introduces the term Broadband Network Gateway (BNG) to differentiate from the legacy ‘BRAS’ term
Transport Mobile RAN 2G, 3G R99, 3G R5, R8 Transport Guaranteed bandwidth, delay and jitter synchronization (frequency and phase)
accuracy inline with Mobile Radio technology
HSI Wholesale Ethernet,
DSL, PON Transport Aggregated bandwidth on ISP level, differentiated services support, with subscriber management at ISP, with L2TP or MPLS VPN transport
Triple Play Wholesale Ethernet,
DSL, PON Transport Aggregated bandwidth on ISP level, differentiated services support, transparent P2P Ethernet transport for unicast services, P2MP Ethernet transport for IPTV
Contribution Video Ethernet, Video HD-SDI Transport Guaranteed bandwidth, delay, jitter , and close to zero or zero loss
Video Service Edge • Implemented on Aggregation Node • Layer-3 MPLS/IP unicast VoD and multicast IPTV
transport for video service distribution
HSI/VoIP Services Edge • Implemented on Centralized BNG • IPoE and PPPoE service transport over 802.1Q and QinQ
interfaces enabled by per subscriber ISG sessions 27
Hybrid (Centralized) Service Edge
DSL Access Node
Access!
PON Access Node
Ethernet Access Node
AggregaPon Network MPLS/IP
Carrier Ethernet Aggregation!
DistribuPon Node
DistribuPon Node
AggregaPon Node
AggregaPon Node
AggregaPon Node
Core Network IP / MPLS
VoD
Content Network
TV SIP
Multiservice Core!
Business
Corporate
ResidenPal
STB
ResidenPal
STB
Business
Corporate
Business
Corporate ResidenPal
STB
Business
Corporate
MPLS/IPoDWDM OpPcal Network
BNG
Ethernet Access Node
BNG
Core
Core
Internet Peering
Core
BSC/RNC
BSC/RNC
HSI/VoIP Service Edge
IP Edge!
Video Service Edge
ASR1000 series:
Up to 64k sessions H-QoS FW, DPI,CGN
MPLS/IP Packet Aggregation for 3play Service Delivery
28
Video Service Edge
Implemented on Centralized Video-BNG
Layer-2 VPLS transport of unicast VoD and multicast IPTV for video service distribution
Centralized Service Edge (with L2 Aggr.)
DSL Access Node
Access!
PON Access Node
Ethernet Access Node
Carrier Ethernet Aggregation!
DistribuPon Node
DistribuPon Node
AggregaPon Node
AggregaPon Node
AggregaPon Node
Core Network IP / MPLS
VoD
Content Network
TV SIP
Multiservice Core!
MPLS/IPoDWDM OpPcal Network
HSI-‐BNG
Ethernet Access Node
Video-‐BNG
Core
Core
Internet Peering
Core
BSC/RNC
BSC/RNC
VFI!
Video Service Edge
HSI/VoIP Service Edge
IP Edge!
Business
Corporate
ResidenPal
STB
ResidenPal
STB
Business
Corporate
Business
Corporate ResidenPal
STB
Business
Corporate
VFI!VFI!
VFI!
HSI/VoIP Services Edge Implemented on Centralized HSI-BNG IPoE and PPPoE service transport over
802.1Q and QinQ interfaces enabled by per subscriber ISG sessions
Distributed Service Edge
3Play Service Edge Implemented on Integrated Edge Node Unicast services (HSI/VoIP/VoD) enabled by IPoE or PPPoE per subscriber ISG sessions Multicast services (IPTV) coexist with ISG sessions Aggregation network implements MPLS/IP for unicast and IP multicast for service transport
MPLS/IP Packet Aggregation for 3play Service Delivery
29
DSL Access Node
Access!
PON Access Node
Ethernet Access Node
AggregaPon Network MPLS/IP
DistribuPon Node
DistribuPon Node
Integrated Edge Node
Integrated Edge Node
Integrated Edge Node
Core Network IP / MPLS
VoD
Content Network
TV SIP
Optional L3VPN Edge!
Multiservice Core!
MPLS/IPoDWDM OpPcal Network
Ethernet Access Node
Core
Core
Internet Peering
Core
BSC/RNC
BSC/RNC
Video/HSI/VoIP Integrated Service Edge
Business
Corporate
ResidenPal
STB
ResidenPal
STB
Business
Corporate
Business
Corporate ResidenPal
STB
Business
Corporate
Carrier Ethernet Aggregation!
C7600
Based on ES+ Up to 48K sessions Limited IPv6 roadmap
ASR9000: Up to 128K+ Sessions Full IPv6 feature set Very good scalability in combined BNG +MSE+CE apps. IOS-XR
The architectures options can be evaluated against the following criteria • Capital Expenditures • Scalability (Bandwidth / Subscriber, Transport, Policy Control)
Operation ‒ Top PE sends “pre-configured” BPDUs (best root) into L2 access network ‒ Access network runs normal MSTP, MSTP is terminated locally on the PE access ports ‒ MSTP TCNs trigger VPLS MAC Flush + Withdraw
‒ MST instances have per port local significance – greatly improves scalability ‒ Only subset of functionality needed for REP Access Gateway
Benefits ‒ Seamless integration with any L2 access network or node running MSTP, full standard compliance
‒ Inherent scalability and faster L2 convergence due to local Rapid STP behaviour
Network-based HA Example 2/3 H-VPLS Spoke (P2P PW) – coupled & “one-way
45
The remote VFI Routers don’t have to run MC-LAG. If it run MC-LAG, it need to be in “decouple mode”
Bundle/POA status decide the PW status. On active POA, it will send active PW status on its primary PW and standby status on its backup PW. On the standby POA, it will send standby PW status on both of its primary and backup PW
Network-based HA Example 3/3 L3 Service – IRB/BVI(asr9k) or SVI(7600) decoupled mode
46
Configure L2 sub-interface 2 PW between two POA. Both L2 sub-interface and L2 PW are in the same bridge-domain. Configure IRB/BVI for the bridge-domain for the L3 service
L3 features like HSRP, VRRP, routing, etc are configured under BVI interface
BVI interfaces are up on both POA regardless of the bundle status
Bundle failover only impact the bundle itself. BVI and related L3 topology is not aware fast L3 convergence
Asr9k - IRB/BVI feature supported from 4.0.1 release
On the bundle, and then configure LDHD configuration option 1: DHD can have default IP gateway pointing to HSRP/VRRP virtual IP address. POA need to configure HSRP/VRRP under BVI interface
Option 2: DHD can also run IGP with both POA. Routing session will be up with both POAs
Bundle/POA failover won’t cause the L3 topology change
“Simplicity is prerequisite for reliability” Edsger Dijkstra
"Simplicity is the ultimate sophistication" Leonardo da Vinci
Kiss: Keep It Simple Straighforward
48
IP FRR: The Principle of Simplicity
Gains
IP Fast ReRoute Theory & Terminology
49
Path: Outgoing interface and next hop Backup: an outgoing interface/nhop which is used to replace another one that went
down. It can be: ‒ another primary ECMP nhop ‒ a secondary LFA routing path
LFA: Loop-Free Alternate ‒ N is an LFA for S’s primary path to D via F if ND < NS + SD ‒ Node-protecting LFA if: ND < NF + FD ‒ Downstream LFA if: ND < SD
Computation of LFA occurs after calculating the primary path, therefore IGP FC performance is not affected
Integrated with LDP Because LFA is precomputed and installed in the FIB, it provides deterministic
protection(<50ms)
IP Fast ReRoute Per-Prefix LFA Algorithm
50
For IGP route D1, S’s primary path is link SF. S checks for each neighbor N (<>F) whether ND1 < NS + SD1 (Eq1) ‒ “does the path from the neighbor to D1 avoid me?” ‒ If so, it is a loop-free alternate (LFA) to my primary path to D1 ‒ C is an LFA for D1, E is an LFA for D2
Scale - Interconnect 100k Access nodes through an MPLS domain Resilience - < 50msec convergence as often as possible Simplicity - Operation of big MPLS networks is often considered difficult
Each IGP area has routes for that area only plus routes to core ABRs ( ~1k prefixes) LDP labels are used to traverse each area and reach core ABRs BGP labels are used by PEs and ABRs to reach PEs in remote areas Service (e.g. PW) labels are used by PEs
IGP/LDP Label
BGP3107 Label
Service Label
iBGP3107 PE21 and D2 via ABR21
iBGP3107 PE21 and D2 via ABR11
L1 L2 L1
D1
PE11
PE12
ABR11
ABR12
ABR21
ABR22
PE21
PE22
D2
NH: ABR21
Label: L1
NH: ABR11
Label: L2
L2 L2 L1 L1
BGP Fast Reroute (BGP FRR)—enables BGP to use alternate paths within sub-seconds after a failure of the primary or active paths
Convergence is driven from the routing protocols updating the RIB and FIB one prefix at a time - Convergence times directly proportional to the number of affected prefixes
With backup paths
Paths in RIB/FIB available for immediate use
Predictable and constant convergence time independent of number of prefixes