Delivering Carrier Ethernet in the Core and Metro

1
Delivering Carrier Ethernet in the Core and Metro
Ananda Rajagopal, Product Line Manager, Foundry
Networks Clayton Wagar, Solutions Marketing,
Alcatel-Lucent Matt Hayes, Mgr Network
Engineering, Cox Communications Peter Green,
Metro Ethernet Applications Cable MSO, Nortel
Networks
2
Problems in the MSO Environment for L2VPN?

• P So many technologies to choose from
• P OAM Troubleshooting Tools? How can we manage
  the network?
• P MTU Issues mismatch or encapsulation
  overhead not accounted for
• P VLAN Mapping Issues
• P L2 Loop Mitigation
• P MAC Learning / Flooding

3
How Can MEF Carrier Ethernet Solve these
problems

• MEF is Technology Agnostic
• PBT, VPLS/MPLS, VLAN, EoSONET, etc.
• 5 Attributes of Carrier Ethernet
• Service Definition
• Scalability
• High Availability
• Service Management
• QoS

4
Potential Pitfalls Carrier Ethernet Can Help
Prevent

• MTU always an issue with so many encapsulations
  involved in L2VPN
• Tunneling of L2CP (CDP, STP, VTP, etc.)

5
Need For Standardized Services

• Standardize expected behavior
• For a Service Type
• At a UNI
• Compliance to standardized services verified by
  accompanying abstract test suites
• Service-centric, transport agnostic
• See picture below

APP Layer
Focus of MEFs Carrier-Grade Ethernet Service
Model
ETH Layer
TRAN Layer
6
Implementing Standardized Services

• Several options are available at core and
  aggregation layer
• IEEE 802.1Q
• Provider Bridges IEEE 802.1ad and/or Provider
  Backbone Bridges IEEE 802.1ah
• VPLS (RFCs 4762)
• Ethernet over MPLS Transport (RFC 4448)
• Handling of customer control frames clearly
  defined at UNI
• Tunnel, peer or drop
• Converged infrastructure
• Traffic management parameters are defined as
  service attributes for service type and at UNI

CMTS
To Head-end
Edge QAM

7
Scalability

• Scalability of Carrier Ethernet is particularly
  important in core and aggregation
• From In-Stats 2006 Ethernet Primary Research
• 1 Demand Driver Scalability
• (Business) customers report 20 to 30 capacity
  growth per annum
• (Business) customers rate Ethernets scalability
  as vital to meeting capacity demand
• Chief attractions of using Ethernet
• Linear scaling in bandwidth needs from 10 Mbps to
  10 Gbps
• Rates beyond 10 Gbps possible today using link
  aggregation
• Work underway in standardizing a 100 Gbps
  interface at the IEEE High-Speed Study Group
  (HSSG)
• Configurable traffic policers provide
  fine-grained control of rates offered at a UNI
• As traffic needs continue to expand, only
  Ethernet has the ongoing scalability to keep
  pace!
• Economies of scale by using Ethernet
• Liquid bandwidth ability to upgrade user access
  speed on the fly without doing a truck roll

8
Best Practices Scaling a Layer 2 Service (1)

• MAC addresses and learning
• For point-to-point services (EPL or EVPL), this
  is not an issue since MAC address learning need
  not be done
• For multi-point service (E-LAN), techniques
  depend on the underlying transport mechanism used
• H-VPLS
• IEEE 802.1ah (Provider Backbone Bridges) aka
  MAC-in-MAC
• Limit MAC addresses consumed per E-LAN service
  instance
• MSOs may also use number of MAC addresses per
  E-LAN service to offer tiered rates when selling
  carrier Ethernet service to end-customers
• Number of peers in an E-LAN service
• H-VPLS By separating spoke and hub sites, a full
  mesh of peer connections are no longer needed
• IEEE 802.1ah Backbone bridges only learn
  addresses of other backbone bridges

9
Best Practices Scaling a Layer 2 Service (2)

• Up-sell opportunities
• Selling incremental bandwidth controllable from
  the NOC with the same equipment (simple
  modification to CIR / EIR as defined in MEF 9)
• Managing/Mapping VLANs in a Carrier Ethernet
  network
• S-VID translation as per MEF 6 and IEEE 802.1ad
• Map S-VID of incoming frame to a unique VPLS
  instance in aggregation node
• Map C-VID of incoming frame to a unique VPLS
  instance in aggregation node
• Map S-VID of incoming frame to I-TAG in 802.1ah
• Benefit Avoid need for VLAN coordination
• Preventing broadcast storms
• Split horizon technique for handling unknown
  unicast or broadcast traffic in VPLS (inherent in
  VPLS)
• Spanning tree enhancements
• Using a default B-MAC address when using
  802.1ah
• Take topology into consideration
• Service attribute regarding broadcast handling
  defined in service type

10
Reliable Converged Architecture for
CableReliability and Service Management
Full suite of OAM tools Ping, Traceroute LSP
ping, Service ping, VPLS MAC ping BFD, VCCV IEEE
802.3ah PW Status Signaling
Integrated element and network management
High availability
Converged Residential and Business Services
DOCSIS
IP/MPLS
CMTS
7750 SR
Fiber
High availability
Node Level Resiliency Non-stop routing for ALL
protocols (LDP, OSPF, BGP, IS-IS, multicast,
PIM-SM) Non-Stop Service for ALL services (VPLS,
VLL, IP-VPN, IES, multicast)
Network Level Resiliency MPLS FRR Sub 50 ms
restoration MPLS Standby Secondary End to end
path protection Dual-homing with
MC-LAG Redundant PWs
E-Line and E-LAN CarrierEthernet Services
11
Non-stop Carrier Ethernet Services
MPLS or Ethernet uplinks with lt50ms failover
Metro core with lt50ms link failover for thousands
of services
Hub
Layer 3 IP VPN
Commercial Business Services (E-Line E-LAN)
Layer 2 VPLS VLL
SDB/SO CTL
D2A

• Dedicated Fiber end-points
• DOCSIS end-points
• MEF E-Line
• EPL (Ethernet Private Line)
• EVPL (Ethernet Virtual Private Line)
• MEF E-LAN (VPLS)
• VPRN (IETF RFC-4364 IP VPNs)

Off Air
EQAM
Video Server
VOD CTL
ATV
CMTS
Ad Insertion
Satellite
A2D
STB Ctrl
DTV
IP/MPLS 10GbE Optical Metro Ring Network
VOD Cluster Servers
Hub
VOD
VoIP
HSI
VoIP Gateway
Headend
Hub
12
Ensuring SLA Performance Metrics for Video, VoIP
and Business
Composite Services
OAM Toolkit for rapid trouble shooting
Enterprise
Fast Service Activation Fault notification
Customer Web Portal
Service Aware Management
Maintain SLA Performance MetricsTest Service
Latency, Jitter, Packet Loss and Round-trip Delay
OAM Notification
Customer A
Commercial Business Services
Customer A
Hub
Layer 3 IP VPN
Layer 2 VPLS VLL
Customer A
HeadEnd
SDB/SO CTL
IP/MPLS 10GbE Optical Metro Ring Network
D2A
Customer B
Customer C
EQAM
Video Server
CMTS
Consumer Services
Ad Insertion
STB Ctrl
Service Assurance Test Customer A
13
Example CoS-based Metro Ethernet SLA

• Service Performance defined in MEF 10.1
• E-Line Virtual Private Line Service
• 4 Classes of Service
• CoS determined via 802.1p CoS ID
• Common type of SLA used with CoS-based IP VPNs

Storage Service Provider
UNI
CE
Carrier Ethernet Network
UNI
ISP POP
CE
Internet
UNI
UNI
Point-to-Point EVC
CE
14
Frame Delay and Delay Variation

• Frame Delay
• This is measured as the time taken for service
  frames across the network
• Frame Delay is measured from the arrival of the
  first bit at the ingress UNI to the output of
  the last bit of the egress UNI. I.e. an
  end-to-end measurement as the customer views it.
• Frame Delay Variation
• Frame Delay Variation is therefore the variation
  in this delay for a number of frames. This delay
  is an important factor in the transmission of
  unbuffered video and where variation occurs in
  the millisecond range can affect voice quality.
  For data can cause a number of undesirable
  effects such as perceived frame loss, etc
• Note The term Jitter is not an appropriate term
  to be substituted from Frame Delay Variation
• Note The MEF expresses performance of delay and
  delay variation in percentage terms
• Note For most purposes one way delay (rather
  than round trip delay) is required to establish
  service quality

15
Frame Loss Defined

• Frame loss is a measure of the number of lost
  service frames inside the MEN.
• Frame loss ratio is frames lost / frames
  sent

CE
CE
CE
CE
time
Metro Ethernet
Metro Ethernet
Network
Network
5000 frames in
UNI to UNI
UNI to UNI
4995 frames out
5 frames lost/or received as errored 0.1 Frame
Loss Ratio (5/5000)
16
Simple and effective Traffic Engineering with OAM
enables Ethernet services with CoS

• Traffic Engineering for CoS
• Eliminate dynamic broadcast-learn approach which
  causes unpredictable forwarding paths
• Utilizes a signaling or management system to
  create deterministic forwarding paths
• Separation of data from control plane
• Resource allocation for pinned paths
• Connection-oriented networking
• Maintain source and destination address
  significance end to end to simplify connection
  management
• Enables destination to identify the source and
  observe the assigned traffic profiles
• Utilize Ethernet protection switching (ITU
  SG15/Q9 g.8031)
• OAM Standards to Monitor CoS
• IEEE 802.1ag Partitions network into
  hierarchical administrative domains
• Continuity checks, Loopbacks and Link trace
• ITU Y.1731 Provides a tool set that enable SLA
  verification
• Frame loss, Frame delay and Frame delay variation

Management
Signaling/Control
Forwarding/Data
customer
demarcs
Adapt
Adapt
Service OAM (SID)
UNI
UNI
Link
Link
Trunk OAM
Link OAM
Link OAM
Link OAM
Edge
Edge
Transit
NNI
NNI
Switch
Switch
Switch
Link
Link
17
Summary

• Carrier Ethernet was Built specifically for
  Service Providers to Offer Rich Ethernet Services
• MSO can greatly benefit from adoption of Carrier
  Ethernet Standards
• Core/Metro Ethernet Networks play a critical role
  in delivering Carrier Ethernet Attributes

18
Thank You!