Post on 29-Apr-2018
Light Reading Megatest Cisco IP Video Infrastructure
Carsten Rossenhövel, Managing DirectorEuropean Advanced Networking Test Center
Agenda
Introduction and Project BackgroundCisco and Spirent EquipmentResults: IP Video Service Delivery Network
Converged NetworkVideo Contribution ScenariosDistribution Scenarios
Results: Experience and MonetizationResults: Data CenterConclusion
About EANTC
European Advanced Networking Test CenterProviding independent network quality assurance
since 1991
EANTC Berlin, Germany
Test and certification of network components for manufacturersNetwork design consultancy and proof of concept tests for service providersRequest for Proposal (RFP) support and life cycle testing for large enterprises and government organizations
EANTC’s
Past Light Reading Test Projects
2003-Extreme Networks
2004-Cisco2002-Alcatel
2008/2009 -
Arbor, Cisco, ipoque, Procera
2007-Cisco
Cisco: “Approaching the Zettabyte
Era”
5
Cisco’s Visual Networking Index Research
Video
IP
InternetVideo
Source: Cisco Visual Networking Index: Forecast and Methodology, 2008-2013
Why Was The Test Conducted?
IPTV Landscape matured since our 2007 testHow do vendors support service providers today
… to differentiate from the IPTV competition?… to create profitable advanced IP video services?… to implement scalable networks and data centers?
How much of Cisco’s vision of a medianet∗
is available today?
∗
Cisco refers to a network that is optimized for video as “medianet”
Commissioned
No financial relationship between Cisco and EANTC for this test
ContractR&R,
Veto Rights
Supplier
Collaborate For Implementation
Who Paid?
Cisco Gear
Service Delivery Network:
CRS-1 ASR 9010
7609-S ONS 15454
Data Center:
Nexus 1000V MDS 9509
DCM 9900 Nexus 7010, 5040
Cisco Demo Lab for Video Applications
“CA
ISL”
Lab,
San
Jose
, CA
Test Equipment: Spirent
Communications
260 x GigE
Ports
5 TestCenter Chassis
46 x10GigE Ports
ANUEXGEM
Impairment
ContributionStudio 1
ContributionStudio 2
7600
7600
Regional Head end
Regional VOD Vault
Nexus5000
VOD Streamer
CDS Cache
IPDSLAM STB – CIS40
PCHome GW
IP Camera
(DSL) Consumer
Business
IP Camer a
ISR
ISR
Business
CRS-1
IP/MPLS
CRS-1
CORE
(Cable) Consumer
uBR10K
RFGW-10 STB – CIS40
PCHome GW
IP Camera
Data Center / Head End
Telepresence
IP Video Surveillance
CDS VOD Vault CDS Internet Streamer
DCM
Encoder
ISDP Server
IPTV / VOD
Nexus 7000
Nexus 5000
ASR1K SBC
Nexus 5000
Digital Signage
MDS9509
FC Disk Array
Ops Mgr Media Server
CRS-1
CRS-1
Cisco “Experience Provider”
Design
15454
15454
L2
(DSL) Consumer
IPDSLAM
STB – CIS40
PCHome GW
IP Camera
15454
MPLS/I P
AGG 7600ASR9K
Service Delivery Network Data Center
Video Applications
Agenda: Areas of Testing
Cisco’s IP Video Service Delivery Network Design
Scale of Test –
The Full Deal2 M Video Subscribers
World’s largest IPTV deployment is ~ 1.85 M subscribers (France Telecom)
20,800 Video Surveillance StreamsEvery Starbucks in 49 countries ( 17,000 stores )Every nuclear power plant in the world ( 400 plants, 50 per
plant )
10,400 Digital Signs1,000 Wal-Mart stores with 10 signs eachEvery European airport ( 450 airports, 20 signs each )
3,120 Telepresence
Sessions100 concurrent sessions in each of 30 DJIA companiesDJIA = Dow Jones Industrial Average
340 Gbps
Traffic per PoPBBC iPlayer’s
peak traffic is less than 100Gbps for whole of UK
3 Failures / RecoveriesCRS-1 RP and DRP
Core Layer
3 Failures /Recoveries
AggregationASR9010 RSP
Aggregation Layer
NGN: Route Processor Failover
(RP)
NGN: Route Processor Failover
Results
Key FindingsThe ASR 9010, CRS-1 RP and CRS-1 DRP all showed hitless RP failover (zero lost packets in MPLS network)
Product Pull RP Switchover (ms)
Insert RP Recovery
(ms)
CRS-1 RP 0 0CRS-1 DRP 0 0
ASR9010 RSP 0 0
NGN: Test Setup –
ASR 9010 Fabric QoS
A9
K-8
T-4B
LC0
10G
10G
10G
10G
Legend“Hi-Priority Protected”
VoIP 382Mb/s per 10G
“Hi-Priority Protected”IPTV & PPV 660Mb/s per 10G"Rest of Baseline 8Medianet
Services
Baseline Setup: 1.No Congestion2.LC3 port BW = 7.9G3.LC3 Lin
e Card BW = 31.6G
4.Loss = 0
SwitchFabric
To accessTo core
10G
10G
10G
10G
A9
K-8
T-4B
LC4
10G
10G
10G
10G
A9
K-8
T-4B
LC0
10G
10G
10G
10G
A9
K-8
T-4B
LC2
10G
10G
10G
10G
A9
K-
LC3
ASR9010
Single Device Stress Test: Fabric Output Queuing
1.5
Gbit/s
High Priority Load:1.Add 1.5 Gbit/s per port of
high priority multicast traffic2.Still no Congestion3.LC3 port BW = 9.4 Gbit/s4.LC3 Lin
e Card BW = 37.6 G
5.Loss = 0
Congestio
n
8 Gbit/s
8 Gbit/s
8 Gbit/s
8 Gbit/sCongestion Setup: 1.Add 8 Gbit/s low-priority
unicast
traffic per port2.LC3 port BW = 17.4 Gbit/s3.LC3 line card BW = 69.6 G4.Loss of low-p
riority streams
NGN: ASR 9010 QoS
–
Fabric QoS
Results
Service Type
Frame Loss
Latency (ms) Jitter (ms)Idle Congested Idle Congested
Multicast (IPTV + PPV)
0 0.037 0.112 0.017 0.045
Unicast VoIP 0 0.012 0.072 0.035 0.067
Key FindingsThe ASR 9010’s “A9K-8T-4B” card (8 port 10 GbE) maintains proper prioritization on the line card and fabric level for up to 69.6 Gbit/s (40 Gbit/s non-blocking) egress traffic
NGN: ASR 9010 Mixed Class Throughput
Key FindingsThe ASR 9010 delivered 160 Gbit/s of multicast and unicast traffic per slot (80 Gbit/s in each direction)
ASR 9010 Mixed Class Throughput Results
0123456789
10
1280 1346 1387 1500 1518
Frame Size (Bytes)Th
roug
hput
per
Inte
rfac
e (G
bit/s
)
UnicastMulticast
Single component
Procedure based Realistic frame Unicast to
Video Contribution Network Tests
Cable
Telco
DVB-T/H, ATSC
Sport Events
Studio-to-studio
Live News Gathering
Event site
Studio
Studio
Stadium
Playout /Uplink Center
Studio
Headend
Headend
Headend
IP / MPLS Network IP / MPLS
IP / MPLS
IP / MPLS
IP / MPLS
CONTRIBUTION DISTRIBUTIONPRIMARY
SECONDARY
SDI (uncompressed) at 1080p MPEG-2/1080p MPEG-4
NGN: Video Contribution IP-CBR Video Monitoring Test
Video Source
Input IP-CBR Video Monitoring
Best Quality
Test ControlFor IP-CBR
Video MonitoringDUT for IP-CBR
Video Monitoring
Delay Factor (DF) –
How much time to drain the buffer?
Media Rate Variation (MRV) –
How much do we vary from the configured CBR?
NGN: Video Contribution Delay Factor Comparison Cisco and Spirent
Spirent XGEM induced periodically (Gaussian) distributed delay from 75 to 125 ms
Cisco and Spirent Monitoring of Group 232.0.1.26
15.000
16.000
17.000
18.000
19.000
20.000
21.000
22.000
174 201 237 243 256 263
Intervals
Cis
co D
F an
d Sp
irent
Max
Jitt
er
(mic
rose
cond
s)
Cisco DF (microseconds) Spirent Max Jitter (microseconds)
Key FindingsCisco 7609-S router accurately detects and reports impaired uncompressed video streams using Cisco IP-CBR Performance tested up to 40 Gbit/s per slot
Video Distribution Network Tests
Cable
Telco
DVB-T/H, ATSC
Sport Events
Studio-to-studio
Live News Gathering
Event site
Studio
Studio
Stadium
Playout /Uplink Center
Studio
Headend
Headend
Headend
IP / MPLS Network IP / MPLS
IP / MPLS
IP / MPLS
IP / MPLS
CONTRIBUTION DISTRIBUTIONPRIMARY
SECONDARY
SDI (uncompressed) at 1080p MPEG-2/1080p MPEG-4
Cisco’s Video Distribution Topologies
PIM-SSM Point-to-Multipoint RSVP-TE
MPLS Fast Reroute used for Link Failure tests
Cisco offered two solution designs –IP and MPLS based multicast
NGN: Video Distribution Video Quality Monitoring (VidMon)
Cisco 7609-S with ES+40 cards tested for MPEG-2 video quality monitoring
Based on Media Delivery Index (MDI, RFC 4445) -
Media Loss Ratio (MLR)
Cisco reporting compared to Spirent TestCenter’s Continuity Error (ETSI standard TR 101 209)
asdf
Video SourceInput MDI Video Monitoring
Best QualityTest Control
For MDI Video Monitoring
Video MonitoringPoint-to-Multipoint RSVP-TE
PIM-SSM
MPEG-2 Multicast Transmission
Anue Loss Impairment
Device
First look at Cisco 7600 upcoming software with HW-based inline VidMon: Monitor MPEG-2 stream quality at egress 7600 ES+ cards
NGN Test –
Video Distribution Networks Video Quality Monitoring Test Setup
NGN: Video Distribution VidMon
Test Results
Total uPE1 Card 3 MLR Values
Total Spirent Continuity Errors Difference
Total uPE1 Card 2 MLR Values
Total Spirent Continuity Errors Difference
Run1 1139 1141 2 1181 1183 2Run2 972 973 1 935 924 11Run3 1279 1279 0 1273 1274 1
PIM-SSM Topology
Groups 1-25 Groups 26-50
Total uPE1 Card 3 MLR Values
Total Spirent Continuity Errors Difference
Total uPE1 Card 2 MLR Values
Total Spirent Continuity Errors Difference
Run1 947 948 1 931 931 0Run2 1083 1096 13 1085 1082 3Run3 1286 1284 2 1331 1329 2
Point to multipoint RSVP-TE
Groups 1-25 Groups 26-50
Key FindingsCisco's 7600 routers accurately reported on MPEG-2 MLR on both PIM-SSM and P2MP RSVP-TE topologies
Service Delivery Network Data Center
Video Applications
Agenda
Amazon Application on TVContextual merchandise during pause
Session Shifting AcrossTV, PC, Mobile Device
User Generated Content
Hyper Syndicated Video
Advanced AdvertisingAds based on consumer profile
Personal
Interactive
Social
IP Video Applications Demonstrated
Customizable UI
Web2.0 on TVRSS Feeds, Flicker, Weather widget
Walled Garden Content Cloverleaf.com web content on TV
IP Video Applications Amazon Application on TV
Idea: Associate shopping opportunities with video content on TVTransaction-based monetization opportunity(Hopefully) shop owners will share profit with service providerCisco example: Amazon shopping application on TV
TV Program / Video on Demand movie paused by subscriber.Set-top box generates note about related shopping opportunities:
Subscriber consents to search
Selects ashopping option
and completes purchase
IP Video Applications
Amazon Application on TV
IP Video Applications Session Shifting
Start watching a film on TV, pauseResume watching the same content on a PC where you left off… and then do the same on a mobile phone
Benefits: Enables continuous media experience across devicesCreates synergies for SPs across service offerings
Cisco IP Set-top Box IPN430 MC PC Nokia E61i
5
Vault/Content Acquirer Cisco CDS
Service Router
TV Streamers Internet Streamers
IP Video Applications
Hypersyndicated
Video
Over-the-top (OTT) User Generated Content (UGC)
Service Delivery Network Data Center
Video Applications
Agenda
The Data Center In Detail
VoIP
Server
VVVV
VoIP
Server
VVVV
IPTV/VoD
IPTV/VoD
Telepresence
Nexus 7010
Nexus 7010
CRS-1
Collapsed Peering /Core
CRS-1
Internet
Internet
Nexus 5040
Nexus 1000V
Video Surveillance
Digital Signage
App
OS
App
OS
App
OS
App
OS
MDS 9509
FC Disk Array
App
OS
App
OS
App
OS
App
OS
VMWareESX VI4
Aggregation/AccessNexus 1000V
Data Center Results Summary
Nexus 5040 properly forwards FCoE/Eth traffic
mix from a single port
Video apps run fine; zero frames lost from line rate
traffic between two ports
Unified FabricProviders can make major
and minor Nexus 7010 code upgrades without taking it out of service
0 frames lost (full traffic profile with VDC)
In Service Software Upgrade
Using full medianet
traffic profile, Nexus 5000/7000 recover from link failure
in under one second using Virtual Device Contexts (VDC)
Data Center Resiliency
Nexus 1000V relocates virtual machine with new physical Ethernet port without
operator reconfiguration
Virtual Machine Relocation
Conclusions -
Results
What can we take away?Cisco is working to bring new ways for providers to create revenue streams through new productsCisco is serious about the virtualized data centerCisco’s IP Video Service Delivery Network infrastructure can support the needs of
Video Contribution and Distribution networksProviders offering the full catalog of residential and business services
Thank you for your interest!
For further information, please contact us:
EANTC AG Carsten RossenhövelEinsteinufer
17
D-10587 Berlin /Germany
Phone: +49.30.318 05 95-0Mobile: +49.177.2505721Fax: +49.30.318 05 95-10E-mail: cross@eantc.de
www.eantc.de