Architectures and Technologies for Optimizing SP Video Networks

© 2007 Cisco Systems, Inc. All rights reserved. Cisco Confidential Presentation_ID 1

Architectures and Technologies for Optimizing SP Video networks

Rajesh Rajah Consulting Engineer Cisco Systems


Session Objectives

 At the end of the session, the participants should be able to:

Understand the trends for video in the SP Industry

Provide a high level End-to End system architecture

Understand the possible architectures and technologies for Video transport

Understand of Network-to-Video-layer linkages that enable optimized Video transport

Provide a deep dive on key mechanisms and technologies to enhance and monitor Video quality


How do you get your TV today ?


What is IPTV?

Broadband IP Access Network

Today: xDSL, Cable Modem, FTTx, Carrier Ethernet, Future?: 3G, WiMax, ...

Subscriber IP-STB (Set Top Box)

Analog or Digital TV (increasingly HDTV)

IPTV = IP network delivered TeleVision Today it usually includes:

Broadcast channels/Switched Digital Broadcast (SDB)

Video-on-Demand services (VOD)

Digital Video Recorder services (DVR/PVR)

Interactive TV applications (ITV)


IPTV Architecture – View from space

“Glass to glass” experience


Delivery Networks with IP as Underlying Transport

Super HeadEnd (SHE)

Receive, Encode Mux, Encapsulate

WAN

IPmc

National Content

Local Access

Regional HE/VHO

Content Servers

Local/Regional Content

Content Servers/Portal

Rcv, Enc Mux,Encap, Stream

WAN

Regional HE/VHO

CORE

DISTRIBUTION AGGREGATION ACCESS CORE

MSO-Hub

ILEC-VSO

DISTRIBUTION

HFC NET

Mux EQAM

DSLAM

VQE AGGREGATION

Radio Tower

DVB-H

HSDPA

Local/Regional Content

Content Servers/Portal

Rcv, Enc Mux,Encap,Stream

WAN EVDO

WiMax

Satellite

Local Access

Mobile

Wireline

Cable

Local Access

Local Access XM-, L-, S-, K-Band…


MPEG/UDP/IP MPEG/RTP/UDP/IP

Analog or Digital

Encrypted MPEG

Demodulate and demultiplex TV signals. Local channels include PEG (Public, Educational, Government) channels.

To IP network as unicast streams.

Encrypted MPEG

Analog or Digital

Ad Splicer will take in the multicast stream and insert new ad content and output two streams with the same Multicast address, but different source addresses.

Local Affiliate

Middleware is the ‘brain’ of an IPTV network. It includes: -  Electronic Program Guide -  Entitlement System -  Asset Distribution -  Navigation Server It communicates with all set top boxes

Used by both broadcast and VoD

VoD Servers store video assets. The Middleware with the Entitlement system, Session Manager On demand manager, Policy Server for CAC, and video pump enable the streaming of programs.

Compress and encode one channel programming in MPEG-2 or 4; SD, HD and/or PiP. Output is IP multicast stream.

To IP network as multicast stream.

Encrypted MPEG


  More HD Channels   Massive VoD Libraries   Time Shifted TV   Internet Video   Any Stream to Any Screen   Targeted Advertising   Next Generation User Interfaces   Service Velocity   3DTV

“The vision is to give our customers the ability to watch ANY movie, television show, user generated content or other video that a producer wants to make available On Demand”

– Brian Roberts, CEO Comcast – CES 2008

Next Generation Video Service Trends Driving network and in-home architectures…


IPTV – 2nd Wave   On-net only   TV   Higher service velocity   Business Model: B2C

Evolution to IP Video Unified experience and enhanced monetization

Traditional Cable – 1st Wave   On-net only   TV   Limited service velocity   Business Model: B2C

IP Video – 3rd Wave   On-net or Off-net   TV, PC, mobile   Highest service velocity   Business Model: B2B2C

More Open, More Flexible, More Monetization Opportunities


3rd Wave Drives Infrastructure Requirements

Requirement Internet Content (Hulu, Netflix)

Personal Media (YouTube)

3rd Wave Video (including Time-Shift TV)

Services Live, VoD, Interactive, Social VoD, Interactive, Social Live, Time-shift, VoD,

Interactive, CDN Ready

Usage / Devices M Copies : N Subs PC, some mobile

1 Copy : N Subs PC, some mobile

1 Copy : N Subs STB, PC, Mobile

Ingest Feeds Scale / Performance

10s, Non real-time

1,000s, Non real-time

100s, Real-time and Non real time

Storage Scale / Resiliency

10-20K Titles, 10s of Terabytes, Med Resiliency

100M+ Titles Petabytes,

Low Resiliency

100K Titles 100s of Terabytes High Resiliency

Ingest : Playout 1 : 10,000s 1 : < 10 1 : 10,000s

Streams Scale 10,000s Millions 100,000s

Latency Tolerance High (secs) High (secs) Low (<1 sec)

File Formats / Protocols HTTP, MS, Adobe Adaptive Emerging HTTP, MS, Adobe MPEG, H.264, Internet Content

Ready

File Sizes, Caching Benefits

Small to Med, High Caching

Small, Low Caching

Large, High Caching


CPE / So(ware / UI / Apps

• Home Gateway • STBs • PCs • Game Consoles • Mobile Phones

Content Ingest and Transport Edge Network

• IP Edge, QAM and HFC • FTTH • xDSL • On-‐Net and Off-‐Net

Content Delivery Network

• Library Server • Caching Gateway • Internet Streamer

Encoding

• H.264 Encoding • MP4 Wrapping

Video Datacenter

Security

• DRM • License Servers • Security OperaVons

Backoffice

• Billing • EnVtlement

ApplicaVon Servers

• RUI HosVng • ApplicaVon Services

Service PlaXorm

• Session and Resource Management • Metadata • Content Management • AdverVsing

Linear /SDV

• Splicing • Grooming

Unified CompuVng PlaXorm

IP Video Solution – 3rd Wave High Level Functional Areas


CPE / So(ware / UI / Apps Content Ingest and Transport Edge Network (IP Edge, QAM and HFC)

Content Delivery Network

Content Library

Cache Nodes

Internet Streamer

STB/PC with player

Internet

CDN CCPH C2

Off-‐Net OpVon

Video Management

HFC

Encoding

H.264 Encoder and MP4 wrapping

Video Datacenter

Security / DRM Backoffice / Billing

BSS/ EnVtlement/

IdenVty

ApplicaVon Servers

Discovery: Navigation

and Selection

Service PlaXorm

ApplicaVon Router

Policy Server

Service Router

Ad Decision System

PATH SRM

DRM

PC with player

Home Network

IPSTB with player

Home Gateway

Mobile Phone

Linear /SDV

Splicer/ Groomer

File-‐based OnDemand Assets and Linear Programs

Unified CompuVng PlaXorm

Game Console

IP Video Solution - 3rd Wave Functional Blocks, Components, and Flows


Connected Home IP Network

News Gathering

Primary Distribution

Secondary Distribution Production

Sport Events

Studio-to-Studio

Video Data Center

IP Network

Post Production

MWP

Direct to Home Headend

Broadband CDN

IP IP

Core Network

Home Gateway

Headend Telco

IP

Cable

IP

Headend

Over the Air

IP

Headend

IP

Broadcast Media Content Delivery Architecture Key Building Blocks

Post Production & Playout Consumption Transport Content Adquisition

& Signal Processing


Video Service Providers: Taxonomy & Characteristics

Uncompressed, Lossless

Very High bit-rate stream: SD (270Mbps), HD (1.5-3Gbps)

P2P and P2MP (unicast and multicast)

P2MP MPLS focused

e.g. BT M&B, RAI

Compressed

Low/moderate bit-rate streams ~ same as or similar to secondary dist

P2P and P2MP (unicast and multicast)

MPLS & IP technology

e.g. Contribution providers, US national backbones

Compressed

Low bit-rate streams: SD (3-4Mbps MPEG2, 2-3Mbps MPEG4), HD (16-20Mbps MPEG2, 6-10Mbps

MPEG4)

P2P for VOD (unicast) & P2MP for IPTV & CATV (multicast)

MPLS & IP technology

e.g. DT, FT, Comcast, …

Stadium Studio

Mobile Studio Fixed

Studio

Final Studio

IP/MPLS Core

IP/MPLS Core

IP/MPLS Core

Home Network

Access and Aggregation

Super Head End (×2)

DCM

National Content Insertion

CDS CDS VOD content distributing to scale

DCM

Head End (×2)

Local Content Insertion

VQE

VSOs (×100s)

Homes × millions

Higher bw streams More end points


Video Transport Services in the SP Video Ecosystem

Consumption

Headend News Gathering

Primary Secondary Distribution Contribution Production

Ingest

Sport Events

Video Data Center

Post Production

Core IP Network

IP

IP

IP IP

Headend

Headend

Direct To Home

Telco

Cable

Mobile

Studio to Studio

Contribution Service

Studio to Studio Uncompressed

Very High bit-rate Unicast and Multicast

Primay Distribution Service

Content origination to Provider Compressed Low to high

Unicast and Multicast

Secondary Distribution Service

Provider to Consumer Compressed

Low to Moderate bit-rate Unicast and Multicast

Increase number of end points

Increase Bandwidth and SLA Requirements


Access Independence

One headend, one IP network Multiple access networks, Multiple screens


Video-to-Network layer Linkages


Unicast, Multicast Performance

and Scalability

Video Service Assurance (QoS, QoE

monitoring etc)

Admission Control Video Service

Bandwidth Management

Visual Quality of Experience (VQE)

Error Repair, RCC

Video Service & Network

Resiliency against failures, DoS attacks

IP Video / IPTV Solution Network to Video layer Linkages

Network Layer

Video Application Layer


Video is very Susceptible to Loss

  Single packet loss may result in an impairment (unlike voice)

  Loss of different packet types result in different types of visual impairment

  QoE is measured subjectively, eyes of the viewer

  General definition for QoE: Impairments/time Mean Time Between the Artefacts

  Common industry benchmark MTBA = 2 hrs or greater No more than 1 error in a 2 hour movie

  Other metrics such as number of support calls may also be important

Slice error

Pixelisation

Ghosting


MPEG: Impact of packet loss

  Impairment depends on which MPEG frames lost I-frame loss will result in a visual impairment

Limiting loss to a single I-frame in the worst case will limit the level of impairment

Detailed paper at http://www.employees.org/~jevans/videopaper/videopaper.html


What is the most efficient way to control loss? Cost / Complexity Tradeoff

Causes of packet loss:   Excess Delay

Prevent with QoS (i.e., Diffserv)

  Congestion Prevented with Capacity planning, QoS and CAC

  PHY-Layer Errors (in the Core) Insignificant compared to losses due to network failures

  Network Reconvergence Reduce with high availability (HA) techniques and smart engineering

Number of possible approaches, or combinations of approaches.

Loss (Impairments/Time)

Cos

t an

d C

ompl

exity

Re-engineering Required

Pote

ntia

l Ove

r-

Engi

neer

ing Viable-

Engineering

Range of viable engineering options may vary by type of video distribution, service or content


Services Comparison and Requirements Services/ Attributes Broadcast Video Video-on-Demand

(VoD) Internet Data

Transport Multicast Unicast Unicast

Service Separation

Common Video VLAN termination on the U-PE.

IGMP/PIM-based multicast control flow

Common Video VLAN termination on the U-PE. L3 routing between VoD

server and U-PE

VLAN-per-DSLAM for Internet subscriber. L2

Point-to-point Pseudowire from U-PE

to BRAS

Convergence OSPF FC, BFD, Multicast

FC, MPLS TE FRR (Routed PW)

OSPF FC, BFD, MPLS TE FRR OSPF FC, BFD, MPLS

TE FRR

Addressing Private IP addressing Private IP addressing Public/Private IP addr

CPE STB STB PC/Laptop

Access control IGMP profiles/white-lists Middleware/VoD server BRAS

Admission control

IGMP state limits Off-path, RSVP-based

On-path CAC, or Integrated CAC

BRAS


Services Comparison and Requirements - continued

Services/Attributes Broadcast Video Video-on-Demand

(VoD) Internet Data

QoS Priority Separate Video Queue with

Higher priority than VoD

Separate Video Queue with Higher priority than

VoD Best effort

Acceptable Packet drop rate

10-6 (one artifact per 2-hr movie)

10-6 (one artifact per 2-hr movie) NA

Latency (RTT) requirements

<200ms <200ms NA

Jitter requirements

<50ms <50ms NA

QoS WRED No No Yes


  The Primary Technology Challenges are common across Distribution and Contribution

1. Basic transport How to shift the packets … IP or MPLS, native or VPN?

2. Video service SLA How to ensure that the IP / MPLS network delivers the required SLAs Number of potential deployment models and technology approaches Specific focus on controlling loss

Ultimate Goal: Lossless Transport

3.  Service Monitoring and Management How to verify that the IP network is delivering the required SLAs for video, and to identify problem areas

Video/IPTV Optimized Transport System Primary challenges


Transport options – IP/MPLS   For non-multicast traffic and point to point feeds:

Native IP or MPLS. L3VPN, P2P TE, etc

  For multicast, multipoint topologies: –  IP

–  Native (PIM SSM)

–  mVPN

–  LSM (Label Switched Multicast)

–  P2MP TE global

–  PW over P2MP TE

–  mLDP

•  mLDP global

•  mLDP + mVPN

IP

MPLS (LSM)

Multicast

mVPN MLDP

P2MP TE

mVPN


Video Contribution Secondary Distribution

Managed Enterprise mVPN

PIM mode SSM only SSM only SM and SSM Sources per multicast group

1 or 2 1 or 2 1 or 2

Multicast Group scale < 1000 < 1000 100s (S, G) per VPN; 100s of VPNs

Receivers per Group <10 Millions 100s of sites; potentially 1000s

Multicast Tree dynamism 100s of new trees per day; trees static once

established

Static trees Trees are dynamic; joins and leaves may impact core

Admission control and Bandwidth Reservation

Yes (time limited reservations)

No No

Fast ReRoute Yes Yes Yes Offload routing Yes No No Path diversity Yes Yes Yes mVPN requirement ? For wholesale

services Yes

p2mp or mp2mp? p2mp p2mp mp2mp

26

Requirements Comparisons for Multicast Based Services running on a Converged IP network

© 2007 Cisco Systems, Inc. All rights reserved. Cisco Confidential Presentation_ID 27 © 2008 Cisco Systems, Inc. All rights reserved. Cisco Confidential C25-452149-02

Characteristic Plain IP Multicast

p2mp MPLS TE mLDP

Convergence < ~500ms ~50ms < ~1s Offload routing

IGP metric based traffic engineering


Path separation

MoFRR or MTR

MoFRR or MTR

Admission control and bw reservation

RSVP

Scalable mp2mp MVPN

Mapping of Multicast Service Requirements to p2mp technology choices


PIM Source Specific Mode (SSM)

Result: Shortest path tree rooted at the source, with no shared tree.

Middleware

STB

B A C D

F E

Encoder


Advantages of SSM

  Very Simple – Easy to implement, maintain & troubleshoot No RP/MSDP configs No SPTswitchover/thresholds Simpler control plane between independent PIM domains

  More Secure Sources are known in advance Only one source can send to the SSM channel Prevention of DOS attacks from unwanted sources

  More Scalable and Flexible Support for both IPv4 and IPv6 addresses SSM for IGMPv3 clients, SSM-Mapping for IGMPv2 clients Flexibility for Static or DNS-based Mapping in case of SSM Mapping Dissimilar content sources can use same group without fear of interfering with each other (although not recommended for IPTV deployment)


Access Aggregation

End-to-end protocol view – Layer3 Agg

STB Home Gateway

Eg: DSLAM PE-AGG

Core Distribution / regional

Home Network Video Headend

Same choices for all access technologies Different by access technology

PIM-SSM (S,G) joins IGMP membership

IGMP Proxy

IGMP snooping

IGMP: {Limits} {Static-fwd} PIM-SSM PIM-SSM

L3 Transport Options in clouds: Native: PIM-SSM or MVPN/SSM MPLS: LSM / mLDP RSVP-TE Opt.

Source Redundancy

IGMP

PIM-SSM

Video Stream


Access Aggregation

End-to-end protocol view digital (non DOCSIS) cable

Cable STB HFC PE-AGG





IGMP snooping

IGMP: {Limits} {Static-fwd} PIM-SSM PIM-SSM


Source Redundancy PIM-SSM

Video Stream

eQAM HFC


Access Aggregation

End-to-end protocol view – Layer2 Agg

STB Home Gateway

Eg: DSLAM PE-AGG





IGMP Proxy

IGMP snooping

IGMP: {Limits} {Static-fwd} PIM-SSM


Source Redundancy

IGMP

PIM-SSM

Video Stream

L2 access

IGMP snooping


Network Resiliency



Fast Convergence - reduces affect of link outage (~ 500ms)

  Implementation and protocol optimisations   Delivers sub second convergence times for unicast (OSPF, ISIS, BGP)

and multicast (PIM)   Available on all Cisco core and edge platforms   Lowest bandwidth requirements in working and failure case   Lowest solution cost and complexity   Is not hitless – will result in a visible artifact to the end users

Core Distribution (DCM)

Edge Distribution (DCM or VQE)

Primary Stream

Rerouted Primary��Stream

Video Source

Video Receivers

X


Multicast-only Fast Reroute (MoFRR)   MoFRR provides the capability to instantiate resilient

multicast trees for the same content If receive IGMP or PIM join on downlink and have multiple

paths to source send joins on two paths Utilize IGP Link-State database and knowledge of how

networks are designed to ensure streams are path diverse Feed connected receivers from only one of the two received

streams

Monitor the health of the primary stream and upon failure, use the secondary

  A simple approach from a design and deployment and operations perspective

  MoFRR depends on natural spatial diversity of large networks, disjointed physical topology with dual edge to dual core

  Can be used for both loss and lossless approaches and be implemented in the network or on the video end system

= Receiver

= IGMP Join

= PIM Join

= Source


Mapping of Multicast Service Requirements to p2mp technology choices

36 © 2008 Cisco Systems, Inc. All rights reserved. Cisco Confidential C25-452149-02

Characteristic Plain IP Multicast

p2mp MPLS TE mLDP

Convergence < ~1s ~50ms < ~1s Offload routing



Path separation

MoFRR or MTR

MoFRR or MTR

Admission control and bw reservation

RSVP

Scalable mp2mp MVPN


Towards Lossless Video/IPTV Transport: Deployment Scenarios

MPLS TE FRR + FEC or TR

MTR + Live / Live

Fast Convergence +

FEC or TR

MoFRR + Live / Live

Fast Convergence

MPLS TE FRR

MoFRR

TE + Live / Live


Towards Lossless Video/IPTV Transport: Deployment Scenarios

MTR + Live / Live

MoFRR + Live / Live

Fast Convergence

MPLS TE FRR

MoFRR

TE + Live / Live

Recommended approach where some loss is tolerable and topology does not support MoFRR •  Lowest bandwidth

used in working and failure cases

•  Lowest solution cost and complexity

•  Constrained impact of network failures on video

Recommended approach where some loss is tolerable and topology supports MoFRR •  Lowest bandwidth

used in working and failure cases

•  Lowest solution cost and complexity

•  Constrained impact of network failures on video

MPLS TE FRR + FEC or TR

Fast Convergence +

FEC or TR

Recommended where lossless approach is required and topology supports path diversity with MoFRR •  Lowest bandwidth

used in failure cases •  Low solution cost

and complexity •  Does not apply to

all topologies

Options where a lossless solution is required and the topology does not support path diversity with MoFRR


IPv4 and IPv6 Multicast Comparison Service IPv4 Solution IPv6 Solution

Addressing Range 32-bit, Class D 128-bit (112-bit Group)

Routing Protocol Independent, All IGPs and MBGP

Protocol Independent, All IGPs and MBGP with v6

mcast SAFI

Forwarding PIM-DM, PIM-SM, PIM-SSM, PIM-bidir

PIM-SM, PIM-SSM, PIM-bidir

Group Management IGMPv1, v2, v3 MLDv1, v2

Domain Control Boundary, Border Scope Identifier

Interdomain Solutions MSDP across

Independent PIM Domains

Single RP within Globally Shared Domains


Multicast Feature Recommendations Features / Platform Core

(N-PE/PE) Aggregation

(PE-AGG if L2 U-PE)

Aggregation (PE-AGG if L3 U-

PE) Access

(Layer3 U-PE)

Access (Layer2 U-

PE) PIM Sparse Mode

PIM SSM Mapping (Static or DNS)

Multicast Loadbalancing

PIM Fast Hello

RPF Tuning

IGMPv2 Join/Leave

IGMP Snooping

IGMP Fast Leave

IGMP Tuning

ARP Timeout Tuning

(Optional) IGMP Static Joins

Multicast HA


Multicast Feature Recommendations

Features / Platform VHE (7600)

DSLAM Residential Gateway (RG)

STB

PIM Sparse Mode

PIM SSM Mapping (Static or DNS)

Multicast Loadbalancing

PIM Fast Hello

RPF Tuning

IGMPv2 Join/Leave

IGMP Snooping

IGMP Fast Leave

IGMP Tuning

ARP Timeout Tuning

(Optional) IGMP Static Joins

Multicast HA


Quality of Service



U-PE

Internet

GE Ring

SONET/SDH Ring

Access Access Core Edge Edge Aggregation

Hub & Spoke

Hub & Spoke

Enterprise A

Enterprise B

Enterprise B

10/100/ 1000 Mpbs

Enterprise A

10/100/ 1000 Mpbs

10/100/ 1000 Mpbs

U-PE

U-PE

U-PE

N-PE

N-PE

N-PE

PE-AGG P

P P

10/100/ 1000 Mpbs

CE CE

•  Classification •  Policing •  Marking •  Egress Queuing

Access

•  Egress Queuing

Aggregation

•  Marking •  Traffic Shaping

CE

•  Egress Queuing

Edge

•  Egress Queuing •  Congestion Avoidance

N-PE


General QoS Guidelines

 Do not mix UDP & TCP traffic in the same class

 Do not mix Voice & Video traffic in the same class

 Per-subscriber SLA for Voice and Data applications

 Per-subscriber SLA not applicable for Video/IPTV

 Over-the-top (Internet) Video traffic to be treated as best-effort traffic

  If Dual Priority queue is supported, then highest priority is for Voice traffic. (Selective) Broadcast Video traffic may be mapped to the lower priority in the Dual PQ.


QoS Guidelines for Video   Network SLAs

Delay: not critical. Most applications are unaffected Jitter: not critical. IP-STBs can buffer 200 msec Packet-loss: critical. Packet loss rate < 10-6 (one noticeable artifact per hour of

streaming @ 4Mbps ). 1 video packet lost may lead to >500 ms of visible artifacts.

  Packet loss due to queue drops by bursts at aggregation points from multiple sources (also number of hops, link occupation)

  Queue depth sizing using probability analysis, so packet loss rate (e.g. 10-6) is below target

  Single or Separate Video queue for Broadcast Video and VoD based on BW requirements, No. of Queues, CBWFQ/WRR, & No. of traffic classes

  Disable WRED for Video queue   Priority of Broadcast Video traffic higher than VoD traffic   Usually Broadcast Video traffic is not over-subscribed   Use VoD CAC during Insufficient Bandwidth conditions


Video optimised Diffserv Schedulers

  Cisco leads the industry in the development and support of multi-priority schedulers implementations

  Enables differentiation between premium services, requiring bounded delays

B

R

Policer

RED

Scheduler

Bandwidth queue

Strict priority queue

Bandwidth queue

Bandwidth queue

Classifier Per-class policy

RED

Tail Drop

B

R

Policer Classifier

EF #1

Tail Drop

AF #1

EF #2

AF #n


Video optimised Diffserv Schedulers

  With Cisco’s optimised IP Diffserv implementations, worst-case per hop delays <<1ms for high-speed links

  End-to-end jitter of <1ms is realiseable today with Cisco’s video optimised products

References:   Clarence Filsfils and John Evans, "Deploying Diffserv in IP/MPLS Backbone

Networks for Tight SLA Control", IEEE Internet Computing*, vol. 9, no. 1, January 2005, pp. 58-65

http://www.cisco.com/en/US/prod/collateral/routers/ps167/prod_white_paper0900aecd802232cd.pdf

  John Evans, Clarence Filsfils, “Deploying IP and MPLS QoS for Multiservice Networks: Theory and Practice”, Morgan Kaufmann, ISBN 0-123-70549-5.


Service Availability

  Network availability is the fraction of time that network connectivity is available between a network ingress point and a network egress point.

  For video, however, simply having connectivity is not enough, hence service availability is often a more meaningful metric.

  Service availability is a compound metric, defined as the fraction of time the service is available between a specified ingress point and a specified egress point within the bounds of the other defined SLA metrics for the service, e.g. delay, jitter, and loss.


Five 9s Availability Five 9s availability assured through   Selecting carrier class network elements with high MTBF and low MTTR   Ensuring that the network design is resilient with no single points of failure (links, nodes

or shared risks), employing redundancy in both network elements and links.   Using IP and MPLS fast convergence and fast reroute technologies, with fast failure

detection techniques (e.g. IPoDWDM) to minimise packet loss from network element failures

  Employing high-availability techniques (e.g. NSF, SSO, ISSU) to minimise the impact from route processors upgrades or failures.

  Using Diffserv QOS, admission control and capacity planning to ensure that the SLA requirements can be met

  Using transport and application level approaches to recover from any loss experienced, and hence provide lossless transport

  Use a “closely coupled” service management solution, to rapidly isolate and identify service impacting faults when they occur.


Traffic Class Core /Edge/ Aggregation Access UNI

MPLS/IP Ethernet DSL, ETTX DSL WiMAX

PHB DSCP MPLS EXP 802.1P 802.1P ATM 802.16

Control Protocols Network Management

AF 48 6 (6) (6) VBR-nrt nrtPS

Residential Voice Business Real-time

EF 46 5 5 5 VBR-rt rtPS

Residential TV and VoD AF 32 4 4 and 3 4 VBR-nrt

NA

Business Critical In Contract Business Critical Out of Contract

AF 16 8

2 1

2 and 1 2 1

VBR-nrt nrtPS

Residential HSI Business Best Effort

BE 0 0 0 0 UBR Best Effort

IPTV DiffServ QOS Domain Example


Class EXP % Bandwidth

Application

Traffic Classes in an IPTV Network

5

Control

Real Time

Business 2 (in-profile) 1 (out-profile)

20

X

IPTV Video 40 4 (Broadcast) 3 (VoD)

Best Effort 13 0

6

25

2 Routing Protocols, BGP, LDP

LLQ for Voice over IP

Delay sensitive business application, video conferencing

Telnet, SAP access, Email

Internet Access

Example


QoS Classes to Queue Mapping Example


IPTV QoS Design Traffic Class

Cos/ Prec

DSCP 1p3q 1p3q 1p3q

6500/7600 1p3q8t/1p7q8t

GSR/ 7600 OSM

SP Control 6 48 P (Q4) P (Q4) P (Q1) P/Q7T1 CBWFQ

Realtime/ Voice

5 40 P (Q4) P (Q4) P (Q1) P LLQ

IPTV – Broadcast

Video

4 32 Q3 Q3 Q4T2 Q3T2/Q3T2 CBWFQ

IPTV - VoD 3 24 Q3 Q3 Q4T1 Q3T1 /Q3T1 CBWFQ

Business In-contract


Business Out-of-contract


Best effort/ Internet


Example


Resiliency & High-Availability



Resiliency/High Availability (HA)  Device/component level

Dual RP (Non-Stop Forwarding/SSO)

Multiple links (Load-balancing across multiple links)

“Fix” Single point of failure conditions (edge card, router, link, source etc)

 Multicast convergence Unicast Convergence

Multicast Fast Convergence

 Multicast Source redundancy Anycast

Prioritycast

Path redundancy (using duplicate streams)


Multicast Convergence Elements

MCvg = T∆t + U∆t + N(RPF∆t + JP∆t) MCvg = Multicast Convergence Time T∆t = Topology Change Detection Time U∆t = Unicast Convergence Time N = Number of Multicast State Entries

RPF∆t = Reverse Path Forward Application Time JP∆t = Join/Prune Message Processing Time

Convergence time T = T1+T2+T3+T4+T5


Elements of Convergence.. Fast Failure detection

  Loss-of-signal (LOS) - SONET/POS, GigE LOS alarms

  Bidirectional Forwarding Detection (BFD) - IETF  Protocol-independent method to detect control/data-plane “liveliness” between two peer systems using hello-like mechanism  Provides sub-second failure detection

Unicast Routing Protocol Convergence   Non-stop Forwarding (NSF), Graceful Restart   IGP Fast Convergence

 Tuning of IGP timers (LSA gen, Throttling, backoff etc)   Incremental SPF (iSPF)   IP Event Dampening   Enable higher priority (route-tagging) for Video Headend Prefixes

  BGP convergence optimization  BGP Update Packing, PMTU discovery etc

Before BGP Convergence Optimization

With BGP Convergence Optimization

0%

20%

40%

60%

80%

100%


…Elements of Convergence

 Multicast Sub-second convergence Set of IOS CLI for the following

Millisecond timers for PIM hello messages

Rapid, triggered RPF interface calculations

Improved IGMP and PIM state maintenance


Redundancy models

  Dual streams (1+1 streams) Let the receiver decide which one to take More applicable in cable vs. DSL/FTTH

  Heartbeat Active sends periodic hello to standby (muted) source

  Anycast Source Two (or more) sources actively sending with same origin IP address Network decides which one to use using its metrics Disaster-recovery and redundant headend applications IGMPv3 or IGMPv2

  Receiver driven Same group with two sources. STB decides which one to join using IGMPv3 Requires IGMPv3 support on STB


Source Redundancy (Duplicate Streams)

S1,G S2,G

I’m responsible for dropping duplicate packets

STB


Source Redundancy (Server Heartbeat)

S1,G S2,G

I will only receive one stream at a time

STB


Source Redundancy (Server Heartbeat)

S1,G S2,G

I will only receive one stream at a time

X

STB


Regional Backbone

Regional Backbone

Service Edge National Backbone Source Residence Regional Backbone

Regional Backbone

Primary Source 1

P

P

Secondary Source 1

Hea

rtbea

t

Primary Source 2

Secondary Source 2

Hea

rtbea

t

Primary Source 3

Secondary Source 3

Hea

rtbea

t

PE

PE

PE P

P

P

P

P

P

PE

PE

X

PE

PE

Native IP Multicast Video Triple Play Redundancy : Video Source Failure


Source Redundancy (SSM)

S1,G IGMPv3 Report

S1,G Join

S1,G S2,G

I’ll try the Primary source, S1,G.

STB


Source Redundancy (SSM)

S1,G S2,G

S2,G Join

S2,G IGMPv3 Report

It appears the Primary source failed. I’ll switch to the Secondary source, S2,G.

STB

X


Anycast Sources

1.1.1.1 1.1.1.1

IGMP Report

v2 join

I will send join to the nearest 1.1.1.1/32

IGMP Report


v2 join

STB STB


Anycast Sources

1.1.1.1 1.1.1.1


v2 join

STB STB

X


Source Redundancy Anycast/Prioritycast policies

Policies

Anycast: clients connect to the closest instance of redundant IP address

Prioritycast: clients connect to the highest-priority instance of the redundant IP address

  Policy simply determined by routing announcement and routing config

Anycast well understood

Prioritycast: engineer metrics of announcements or use different prefix length.

  No vendor proprietary source sync proto required

  Per program, not only per-source-device failover Use different source address per program

Src B secondary

10.2.3.4/32

Rcvr 2 Rcvr 1

Src A primary

10.2.3.4/31

Example: prioritycast with Prefixlength announcement


Source Redundancy Anycast/Prioritycast benefits

  Sub-second failover possible   Represent program channel as single (S,G)

SSM: single tree, no signaling, ASM: no RPT/SPT

  Move instances “freely” around the network Most simply within IGP area Not good for eg: regional to national encoder failover

  No vendor proprietary source sync proto required   Per program, not only per-source-device failover

Use different source address per program


Anycast-Source with RIPv2 Update

s s/32, m=1

1

redistribute s/32, metric 10

•  The two sources are active and sending •  s/32 routes are generated by both source using RIPv2 updates •  Host routes for anycast source are redistributed into IGP with variable metrics

(optional) •  Network selects source (PIM join messages) based on metric •  Upon video failure, sources withdraw s/32 routes using Poison Reverse

(infinite metric) updates

ENC ADP

s s/32, m=1

2 ENC ADP

redistribute s/32, metric 5 s

s/32, m=16 1

ENC ADP

s s/32, m=1

2 ENC ADP

X


Regional Backbone

Regional Backbone

Service Edge National Backbone Source Residence Regional Backbone

Regional Backbone

Primary Source 1

P

P

Secondary Source 1

Hea

rtbea

t

Primary Source 2

Secondary Source 2

Hea

rtbea

t

Primary Source 3

Secondary Source 3

Hea

rtbea

t

PE

PE P

P

P

P PE

P

PE

PE X P

PE

PE

Native IP Multicast Video Triple Play Redundancy : Source Router Failure


Multicast Group Based : Multi-path Load Splitting

Active Video Server

Standby Video Server

Hea

rtbea

t

R1

R3 (S,G1) (S,G2) (S,G3) (S,G4)

BEFORE

Active Video Server

Standby Video Server

Hea

rtbea

t (S,G1) (S,G2) (S,G3) (S,G4)

Now

Source + Group Based Load Splitting

All Links Efficiently Used!

R1

R3

R4

R5

R2

(S,G1)

(S,G2)

(S,G3)

(S,G4)

(S,G1)

(S,G2)

(S,G3)

(S,G4)

Source Based Load Splitting

Links Unused R4

R5

R2

Hash based on Source Requires unique sources for load splitting

Hash based on S,G


Multicast HA & Convergence High Availability

HA/Convergence features Broadcast Video Traffic Video-on-Demand traffic

Redundant RP, Power supply, Fan tray, Fabric cards

OSPF Fast Convergence

OSPF iSPF

Bidirectional Forwarding (BFD)

P2P MPLS Traffic Engineering (MPLS TE)

Not Applicable

Multicast sub-second convergence

Not Applicable

L2 Pseudowire


Security



Multicast Security..  Protect router/switch CPU (control plane)

Control Plane Policing (CPP) – Policing on router-wide virtual control plane

Hardware Rate-limiters (HRWL mls ratelimiters)

MQC-based (per-interface)

 Enable multicast protocol filtering/setting administrative boundary

Boundary ACL (Filters control/data plane traffic for specified groups using “ip multicast boundary” CLI)

Receive ACL

 Enable spoof prevention MD5 authentication, PIM Neighbor filters


.. Multicast Security

  Prevent Memory (SW) and Hardware (state) overload IGMP, MLD limits /max-groups IP Multicast Route limits (ip multicast limit CLI)

  Allow traffic only from STBs to Video Servers (data-plane filtering)

Generic ACLs (typically on user-facing intefaces/SVIs)   Restrict access to Channels based on User subscription

Offer Tier-based services (Premium, Gold, Silver packages etc) at Network level

Use of IGMP Profile/access-group CLI on a per-interface basis   Network Address Translation (NAT)

Source address NAT Destination/Multicast Group NAT (aka Service Reflection) Useful when Overlapping address space is present, Integrating existing/new networks, etc


Multicast Admission control IGMP/MLD Limit Commands

Malicious IGMP/MLD Reports

Valid Periodic IGMP/MLD Reports

0

max

time

Tota

l M

emor

y U

tiliz

atio

n

t1 t2

time = t1 time = t2

Gasp!

0

unlimited

time

IGM

P/M

LD

Ent

ries

t1 t2 tn tn

Other Processes

IGMP/MLD Table

Memory Resources

What does it do ? • Sets quota on the number of cached

entries in IGMP/MLD tables • Channel Offering Limits in household

•  Denial of Service has been mitigated!

How it works: •  Time = t1, router receives valid

IGMP/MLD Join(s), populates table(s) and allocates required memory

•  Time = t2, router suddenly receives malicious IGMP/MLD Join(s) and table(s) quickly begins to grow

•  Time = tn, all memory resources are exhausted and router is unable to service other processes requesting more memory

•  Now, user sets IGMP/MLD limit


Ethernet Access Security Threats

Subscribers Switches Infrastructure Layer 2 service isolation across switches

L2 Control Protocol Attack (STP, CDP, VTP, etc…)

Man-in-the-Middle attacks on critical management traffic

Non intentional forwarding of traffic between UNI ports

MAC Flooding / Overflow Unauthenticated access to the switch configuration file

DHCP Rogue Server MAC Flooding / Overflow Unconfigured Ports providing network access

IP & MAC Address Spoofing Unicast, multicast, or broadcast storms

Unauthorized network access, junk traffic

ARP Spoofing (Man-in-the-Middle)

Infected users flooding the network / Malicious users attacking the Priority traffic queue

Unauthenticated network access by client devices

Attack targets can be divided into three main categories:


Common Security Recommendations How to Secure the Network Against Attacks

Leading Practice Category Examples Protects Against Threats

Disable Unnecessary Services ICMP redirects, CDP, IP Source Routing Reconnaissance, Denial-of-Service

Control Device Access TACACS+, Radius, Password Encryption Unauthorized Access

Secure Ports and Interfaces Disable unused interfaces, VLAN Pruning Reconnaissance, Denial-of-Service

Secure Routing Infrastructure MD5 Authentication, Route Filters Denial-of-Service

Secure Switching Infrastructure Port Security, Storm Control Denial-of-Service

Control Resource Exhaustion Control Plane Policing (CoPP), Hardware-based Rate Limiters Denial-of-Service

Policy Enforcement uRPF, iACLs IP Spoofing, Denial-of-Service

DSLAM MAC Forced Forwarding, Virtual MACs, DHCP Option 82, IGMP

Whitelist

Reconnaissance, MAC Spoofing, Theft-of-Service


Residential Access Leading Practices How to Secure Users and Services

Goal Features

Subscriber Identification DHCP Option 60, DHCP Option 82

Subscriber Authentication PPPoE or Web Portal (Using Radius)

Subscriber Isolation MAC Forced Forwarding on DSLAM

Private VLAN/PVLAN Edge on Switch

Rogue DHCP Server DHCP Snooping

Prevent MAC/ARP Address Spoofing Virtual MAC Addresses on DSLAM

DHCP Snooping + ARP Inspection on Switch

Prevent Theft of BTV Service IGMP Whitelist on DSLAM

IGMP Profile/Access-group on Switch

IP address spoofing DHCP Snooping + IP Source Guard (IPSG) on Switch

Limiting No. of Channels/IGMP/Multicast states

IGMP State limits/max-groups & Multicast limits on Switch


Layer 2 Leading Practices

Attack Defensive Features/Actions

MAC Attacks (CAM Table Overflow) Port Security, Per VLAN MAC Limiting

Broadcast/Multicast Storm Attacks Storm Control Thresholds

L2PDU DoS Attacks Hardware Rate Limiters, Control Plane Policing, Storm Control Thresholds

VLAN Hopping, DTP Attacks

Disable Auto-trunking, Use Dedicated VLAN-ID for Trunk Ports, Set User Ports to Non-trunking, VLAN 1 Minimization/Pruning, Disable Unused

Ports

DHCP Starvation Attack DHCP Rogue Server Attack

Port Security, DHCP Snooping, VLAN ACLs to block UDP port 68

Spanning Tree Attacks BPDU Guard, Root Guard

Infected users flooding the network / Malicious users attacking the Priority traffic queue

Rate-limiting, Priority policing

ARP Man-in-the-Middle Dynamic ARP Inspection

How to Secure the Network Against Layer 2 Attacks


Infrastructure Security Leading Practices

Security Threats

Man-in-the-Middle attacks on critical management traffic

Unauthenticated access to the switch configuration

Unauthenticated network access by client devices

Unconfigured Ports providing network access

Unauthorized network access, junk traffic

Out-of-Band Management, SNMPv3, SSH, per-command AAA

Password recovery disable

802.1x

UNI Default Port Down

Access Lists


Visual Quality of Experience



Without VQE VQE Enabled

Noisy Last Mile

Improving Cisco IPTV Experience Non-Stop Visual Quality Experience (VQE) Technology

VQE Server

Aggregation Router

Access Node Access Node

• Caches all Video channels • Retransmits lost packets to STB

Visual Quality

Experience (VQE)


Channel change Events Summary

STB MPEG

Network

STB Related to STB implementation

Related to network delays

Related to STB MPEG buffer Not to scale*

STB STB Network STB

User hits channel change on remote

SW starts channel change STB sends IGMP leave (wire), clear old buffers

STB sends IGMP join (wire)

Leave/Join/Network Latency

STB MPEG Buffer

1st UDP packet arrives at STB

SW recognizes UDP pkt

Start filling jitter buffer

Jitter buffer full Wait for arrival of PSI – PAT, PMT, CAT

Wait for arrival of I-frame

STB MPEG buffer processing complete

STB starts decode

Channel change complete

Video/Audio is played

* t=0


Sample Channel change time calculation AVC/H.264 SD on IPTV DSL

Channel Change Latency Factor Device/Location Typical

Latency Cumulative

Latency 1 Send IGMP Leave for channel X STB < 10 ms 2 Send IGMP Join for channel Y STB < 10 ms

3 DSLAM gets Leave for channel X DSLAM/Network < 10 ms

4 DSLAM gets Join for channel Y DSLAM/Network < 10 ms ~ 20 - 40 ms

5 DSLAM stops channel X, and sends Channel Y DSLAM/Network ~ 30 – 50 ms ~ 50 – 90 ms

6 DSL Latency (FEC/Interleave) DSLAM/Network ~ 10 ms ~ 60 - 100 ms 7 Core/Agg Network Latency Router/Network ~ 20 – 60ms ~80 – 160ms 8 De-jitter buffer STB ~ 300 ms ~ 380 - 460 ms 9 Wait for PAT/PMT STB MPEG buffer ~ 125 ms ~ 500 - 580 ms

10 Wait for ECM/CA STB MPEG buffer ~ 125 ms ~ 620 - 700 ms 11 Wait for I-frame STB MPEG buffer ~ 250 ms to 2s ~ 870 ms – 2.7s 12 MPEG buffer STB MPEG buffer ~ 1s to 2s ~ 1.8s – 4.7s

13 Decode STB ~ 50 ms ~ 1.9s – 4.8s


Optimizing Channel change time – Page 1

Device Optimization Factors

Encoder   GOP length tuning   Tuning PAT/PMT intervals (if supported)

Conditional Access   Tuning of ECM intervals (PMT)   Key rotation timeframe

Residential Gateway (RG)

  Tuning IGMP timers   Video-optimized QoS config#

STB   Cache PAT/PMT   Buffer optimization and play-out techniques

# Not a direct contributor to reduce zap time. But, helps reduce response variability and enables better treatment for Video


Optimizing Channel change time – Page 2

Device Optimization Factors

Headend Router   Video-optimized QoS config #(marking, scheduling etc)

Core Network Elements   Secured control plane #(PIM/IGMP limits, Control plane policing, Hardware rate-limiters etc)   Video-optimized QoS config #

Distribution/Aggregation Network Elements

  IGMP static joins for popular channels   Video-optimized QoS config #   Secured control plane #

Access Network Elements (DSLAM/MetroE switch/

PON)

  IGMP Fast/Immediate leave   Tuning IGMP timers (Query time etc)   Explicit IGMP Host tracking (IGMPv3)   Video-optimized QoS config #   Secured control plane #

# Not a direct contributor to reduce zap time. But, helps reduce response variability and enables better treatment for Video


Visual Quality

Experience (VQE)

Cisco IPTV Fast Channel Change Combined VQE Unicast stream & Client Early Channel Change!

Combined Cisco Fast Channel Change: Average: ~0.7 sec Variance: ~0.4 sec

Un-optimized channel change time stats: Average: ~2.2 sec Variance: ~1.2 sec

Access Node

Set-Top Box

Early Channel Start & VQE I-frame burst

VQE Server +

+ Aggregation Router

•  Caches all Video channels •  Bursts Video streams to STB starting with I-frame


Admission Control



Media-aware IP NGN Video Call Admission Control (CAC)

Video Streams

2 VoD Streams—4Mbps Each

Video Quality Fantastic Video Quality Suffers (for ALL users)

3 VoD Streams—4Mbps Each

10 Mbps

4 Mbps

4 Mbps

VoD TV

10 Mbps

4 Mbps

4 Mbps

4 Mbps

Gracefully Rejects 3rd VoD Stream

10 Mbps

4 Mbps

4 Mbps

4 Mbps

3 VoD Streams—4Mbps Each with Video CAC

End-2-End Video CAC (RSVP-based)

7600 ASR9000

Video Admission

Control


VoD Request

Policy Server

Channel request

Request Denied/ Accepted

RSVP-CAC

Video on Demand Unicast CAC

VoD Servers

Available Bandwidth Check


Network Call Admission Control Avoiding Congestion Packet Loss

Against a DiffServ prioritized percentage of link bandwidths

IPTV Channel Change

Broadcast Source

Policy Server

Channel request

Request Denied/ Accepted

1 4

2

Multicast CAC

Broadcast TV Multicast CAC

1 4

2

3

3 Available Bandwidth Check


Cisco 7600

Cisco 7600


Carrier Ethernet Aggregation"

Core Network IP / MPLS

BNG

BNG

Edge"

DSL Access Node

Access"

Business

Corporate

Residential

STB

Residential

STB

Business

Corporate

Business

Corporate

Residential

STB

PON Access Node

Aggregation Network

IP

Distribution Node

Distribution Node

Aggregation Node

Aggregation Node

Multiservice Core"

Ethernet Access Node

Ethernet Access Node Aggregation

Node

Aggregation Node MSE

MSE

VoD Controller Entitlement Sys

Session Mgt, EPG

Middleware

RSVP Path 4

eg RTSP 3 1

2

VoD Stream

6

RSVP Resv

5 CAC CAC CAC

Pure On-Path CAC for VoD Synchronisation between RSVP and VoD streaming

Content Network

VoD TV SIP

VoD

VoD


Carrier Ethernet Aggregation"

Core Network IP / MPLS

BNG

BNG

Edge"

DSL Access Node

Access"

Business

Corporate

Residential

STB

Residential

STB

Business

Corporate

Business

Corporate

Residential

STB

PON Access Node

Aggregation Network

IP

Distribution Node

Distribution Node

Aggregation Node

Aggregation Node

Multiservice Core"

Ethernet Access Node

Ethernet Access Node Aggregation

Node

Aggregation Node MSE

MSE

VoD Controller Entitlement Sys

Session Mgt, EPG

Middleware

RSVP Path 4

eg RTSP 3 1

2

Pure On-Path CAC for VoD Synchronisation between RSVP and VoD streaming

Content Network

VoD TV SIP

VoD

VoD

RSVP Resv 5

CAC CAC Reject

RSVP PathErr

6

eg RTSP 7

See draft-ietf-tsvwg-rsvp-proxy-proto


Video Quality Monitoring/Assurance



Video/IPTV Quality Measurements (What Can Go Wrong)

Content Measures Picture Quality, Blocking, Blurring, Visual Noise, Audio Drop-outs

Media Transport Measures PCR Jitter, Pixelization, Sync Loss, Continuity Errors

IP Network Measures Packet Loss, Jitter, Delay

Physical

Visual

Ethernet

IP

UDP

RTP

MPEG-TS

Content

Cont

rol

Control Measures IGMP Latency, RTSP Latency, Channel Zap Time

Error Type

QoE Errors Impacts Customer

QoS Errors Impacts Operator

Control Plane Problem

Video Problem

IP Problem

Problem Area


VidMon is a Family of Metrics

 VidMon does not represent a single metric but rather a family of Metrics.

 Not all Routers have the same capabilities and therefore Metrics will vary across platforms.

 The applicability of a VidMon Metric will differ based on the type of Video being Monitored

 VidMon Metrics can be used independently or used to compliment each other.


The VidMon Metrics

Transport IP UDP RTP FCS UDP Video Payload Content

(MPEG is not the only payload option)

Example Video Packet in over an IP Transport

Metric Applicability

Media Delivery Index (MDI) Measures MPEG2/4 Headers for Loss and Delay

Media Discontinuity Counter (MDC) Measures MPEG2/4 Headers for the number of times Loss was detected.

Media Rate Variation (MRV) Measures IP/UDP Headers for Delivery Variations.

RTP Loss and Jitter Measures RTP Loss and Delay by examining the RTP header

Media Stop Event (MSE) Notification if a monitored flow stops receiving traffic

MPEG Header

MPEG Payload


What is Media Delivery Index (MDI)

 MDI is a metric developed in cooperation between IneoQuest and Cisco

 Presented in RFC-4445

 MDI is a combination of two metrics that are used to measure the networks contribution to video impairements.

 The two MDI metrics are: MDI:MLR – Media Loss Rate : Were any MPEG packets dropped

MDI:DF – What is the buffering requirements for these packets


Understanding MDI:DF (Delay)   Difference between the arrival and drain rates of a media stream.

This is largely based on the arrival of the IP flow.

As such the MDI:DF and MRV:DF will appear the same

  Delay Factor is based more on RFC 3393 than on RFC-4445.   The DF over an interval period represents the buffering required to

handle variations in transmission at a point in the transmission path.

  To calculate delay factor the virtual buffer (VB) maximum measured delay rate has the VB minimum measured delay rate subtracted. This value is divided by the media rate over that measurement interval

DF = [VB(max) – VB(min)]/MR


Understanding MDI:MLR (Loss)   MDI measurement of MLR inherently refers to the ability to detect

loss in the media stream itself representing the magnitude of a loss event.

  In VidMon, MLR is calculated by monitoring discontinuities in the MPEG TS headers of a packet.

  The Continuity Counter (CC) exists in each MPEG header and is a rolling 4 bit counter unique to each program (PID).

IP UDP RTP I EEE

Transport Headers

… … Adaptation Control Field

Continuity Counter … …

Adaptation Control Field

Continuity Counter

Could represent the same or Different Program PID

MPEG Frame IP Payload


Regional Network"

Cisco 7600

Cisco 7600

Backbone"

CRS-1

Headend"

CRS-1

DCM

Hub"CMTS

GQAM /XDQA

Cisco 7600

Regional "Headend"

DCM

DNCS

Hub"CMTS

GQAM /XDQA

Preserving QoE MDI Monitoring

MDI: MDI: MDI: MDI:

Problem Detected!

1) Video quality problem detected.

3) Troubleshoot location where MDI first degrades. 2) Measure Media Delivery Index (MDI) at each router between receiver and source

NOTE: MDI is a combined measure of video quality based on packet loss, jitter, latency

CDS Vault/ Content Acquirer

CDS TV or Internet Streamer


CDS Service Router

Problem Isolated


Regional Network"

Cisco 7600

Cisco 7600

Backbone"

CRS-1

Headend"

CRS-1

DCM

Hub"CMTS

QAM

Cisco 7600

Regional "Headend"

DCM

DNCS

Hub"CMTS

QAM

MDI: MDI: MDI: MDI:

Problem Detected!

1) Video quality problem detected.

3) Troubleshoot location where MDI first degrades. 2) Measure Media Delivery Index (MDI) at each router between receiver and source

4) Correct problem and restore video quality.

MDI: MDI:

Problem Solved!

NOTE: MDI is a combined measure of video quality based on packet loss, jitter, latency

CDS Vault/ Content Acquirer



CDS Service Router

Problem Isolated

Preserving QoE MDI Monitoring


Media Rate Variation: MRV

  Some platforms can not measure into the media payload of an IP packet to calculate medial loss.

  Some payload types, such as SDI, HD-SDI are not candidates for a metric such as MDI.

  An alternative approach is to measure loss as a function of the L3/L4 header.

  For Constant Bitrate Flows (CBR) a normalized bit arrival rate can be created based on the known media arrival rate.

  The Video flow is monitored for variations in the arrival rates which represent perturbations caused by excessive delay or loss in the media flow.


Measure CBR Flow Arrival Patterns

(Keohane, 2009)

Normal Case

Error Case


RTP Loss & Delay

  RTP headers can be use in the delivery of video media in an IP network.

  RTP headers include a sequence number which can be used to track loss and a timestamp that can be used to calculate delay.

  RTP would likely not be reported as an MDI metric since it represents discrete measurements.

IP UDP RTP I EEE

(Keohane, 2009) Transport Headers MPEG

Headers MPEG Payload


Market for RTP Measurements

 RTP is an ideal candidate for measuring loss in IP transport.

 RTP is independent of the Video Media type in the payload

Beneficial in uncompressed video transports and non-MPEG video transports

 RTP is not currently widely deployed in the MSO market while more prevalent in the Wireline market.

Newer Video over DOCSIS IPTV applications will likely be RTP based however we are early in the adoption of that technology.


Key Takeaways   A systems view is increasingly important to architect

networks for SP Video

  Advanced network resiliency mechanisms are available to design lossless Video transport

  Video-layer-to-Network linkages offer significant benefits and differentiation

  Video monitoring (esp. In-line) monitoring is very beneficial to Service providers


Q&A

Questions ?

Architectures and Technologies for Optimizing SP Video Networks

Technology

Transcript of Architectures and Technologies for Optimizing SP Video Networks