Title: Key Factors in Developing and Evaluating an IPTV Business Case
1Key Factors in Developing and Evaluating an
IPTV Business Case
November 2005 Jack Barrett Director Business and
Financial Analysis Juniper Networks
2Objective
- Understand the economics of IPTV services to
support Capital Investment decision making - Answer
- What Impacts Demand?
- What impacts Capital Investment?
- What Impacts Operational Expenses?
3Anatomy of Opex, Capex, ROI/TCO in the Network
Edge (Capex)
Core (Capex)
Facilities (Opex)
Total Costs Of Network?
4Economics of IPTV
- Calculating Traffic and Circuits
- Applications
- Take rates
- Total peak bandwidth
- Network Architecture and Engineering Rules
- Capital Investment
- Architecture and number of POPs
- Line Card Protection
- Efficient Scaling
- Operational Expenses
- Running Costs
- Network Management
- Customer Care
- Remediation Costs
- Pricing Analysis (not covered here)
- Market Based vs. cost plus
- Service Costs and Partnerships
To compare alternative implementation we use the
traffic models capital deployment and cost models
5Application Requirements
- Internet Traffic
- 100K per ADSL2 Connection growing at 2 per Year
- Over-subscription 501
- Concurrent peak usage 50
- Penetration S-curve approaching 75 saturation
value - VoD Traffic
- Bandwidth 3.5 M per growing at 5 per year
- Oversubscription 11
- Concurrent peak usage 12
- Penetration S-curve approaching 35 saturation
value - Broadcast TV
- Bandwidth 3.5 Mbps growing at 5 per year
- 400 channels
- 75 channels being viewed
- Penetration S-curve approaching 40 saturation
value - IP Phone
- Average Bandwidth 40Kbps per second
- Oversubscription 51
- Concurrent Peak Usage 35
6Market Size and Penetration of Applications
Network connections will drive OpEx costs (ie
provisioning, monitoring, etc) and will determine
revenue (Price connections) Each connection
will subscribe to various services with a high
majority choosing VOD. This factor is especially
important over time as new unicast based
applications drive additional traffic volumes.
7Total Traffic Load
This chart shows 12 VOD concurrency driving over
90 of your traffic as non-aggregated downstream
unicast. This should drive a question in your
IPTV economics evaluation as to the value of
aggregation for a small percentage of multicast
traffic.
8Reference NetworkSet the Network Architecture
- GE connectivity from DSLAM to Aggregation and/or
BSR - 10 GigE between Aggregation and BSR
- 10 GigE between BSR and Core or Video Servers
9Converting Bandwidth to CircuitsAssumptions
Impacting Deployment
-
- Circuit Utilization 75
- Percent of traffic to core 98
- Oversubscription 11
- No protection on facilities from DSLAM to
aggregation layer - 11 Facility protection (active and non-active)
from DSLAM to Agg and from Agg to Core
10Circuits Required to Support Demand and
Architectural Design
- Traffic is converted to Circuits based on design
criteria and number of locations - Note that some aggregation can be achieve in
first two years due to lightly loaded DSLAMs - Modeling tool can be used to optimize the
deployment of Aggregation points and BSR
locations by setting the number of each to 1 - The number of chassis when setting locations to 1
generates optimum number of chassis to carry
traffic - Will be a trade-off however between back haul
costs and capital by spreading optimum number of
chassis across network -
11Closer look Average Utilization of DSLAM
Facilities
- Circuit utilization at 25 year 1 makes
aggregation possible in the first year - Requires constant monitoring year 1 to year 5 for
growth (OpEx) - Note modeling tools takes this into account to
forecast optimal number of circuits based on
utilization and traffic rules
12Networking Capital Investment
- At this point, we know how many circuits are
needed based on traffic and number of locations,
so capital can be estimated based on - Network Architecture
- Number of boxes required in solution
- Number of locations for each type of equipment
such as DSLAMs, aggregation, and routing points
and growth - Costs and associated capacities of equipment and
line cards used to terminate the ports - N1 Line Card Redundancy
- Scalability
- Server Placement
- Dependent on Cost structures
- Trade-off between capital, bandwidth and
management costs
13Architecture Choices Distributed vs. Centralized
- Model 1. Centralized with Direct connect to DSLAM
- Will minimize operational expense and scale the
best - Potentially minimizes capital expenses
- Enables option to use existing aggregation
capacity or a mix between Model 1 and 2 to
optimize - Model 2. Centralized with simple Aggregation
Layer - Provides a simpler device to management to
minimize opex - Potentially lowers Capital in cases where there
are low utilized DSLAM or there is an existing
aggregation layer - Model 3. De-Centralized Network Intelligence
- Puts network intelligence in less expensive
equipment throughout network - Potentially increases costs by adding more
complex equipment
Model 3 BSR with split functionality
- L2 Aggregation
- Subscriber Mgmt
- Queuing
- Rate limiting
- BRAS
- Routing intelligence
- Protocol mgmt
14Impact of N1 Line Card Protection
1 1 Redundancy Could cut the effective bandwidth
half or risk reduced service availability
R
R
R
R
R
R
R
R
N1 Has little impact on capacity of equipment
R
R
10G ingress
10G egress
15Impacts on Scalability
- Example With little or no over-subscription an
increase of 10 Gig per POP will - Single Box requires one additional 10 Port GE
Card and 1 additional 10Gig Card on egress - An Aggregation switch doubles the number of cards
needed - If N1 line card protection is not available,
then the number of card doubles again - Potentially this could be a 41 ratio
Added Line cards, N1 protection comes with
chassis requiring no additional line protection
cards
Aggregation Layer Solution
Line cards added in 4 places-plus 4 additional
Line protection cards to support 11 redundancy
16Economics of Operational Expense
- Operational Expense enhances competitiveness and
next to revenue attainment is the most important
factor related to profitability - Operational expenses are driven by
- Labor Costs
- Professional Services
- Site preparation,
- Installation, testing and turn-up
- Network management
- Monitoring
- Upgrades and planned outages
- Fault management and troubleshooting
- Configuration Management
- Customer Support
- Service Provisioning
- Trouble calls
- Add, change and deletes
- Billing
- Running costs such as power, space, maintenance
and transmission - The complexity of these process and number of
times they need to be invoked will depend on
equipment reliability, architecture and number of
platforms
17Network Management DriversIllustrative Example
- 37 more chassis in stalled over 5 years
- 300 more line cards and associated line modules
installed over 5 years - Will have a linear increase in operational
expense in terms of - Space and Power
- Installation
- Test and Turn-up
- Increase number of alarms
- Planned outages
- Monitoring Hours
18Customer Service
- Enable Pre-Provisioning and Self Provisioning
- Minimize calls to customer care work centers
- Ensure easy integration of new services
- Application integration and ease of adding
business services - Bandwidth sharing across applications improves
customer satisfaction and reduces churn - Minimize Devices in the Home
- Avoid multiple devices per service
- Simplified billing and better network reliability
will reduce calls into customer care
19Customer Care can have the largest Impact on
Operational Expenses
- Assumes 12 cost to carrier to provision a
service connection (not including ADSL
connections) and shows the impact of cutting
connection requests in half - In the highest growth years this represents a
12.5M savings
20OpEx and RevenueSecurity and Optimization
- Dual Firewalls for strong security with
continuous uptime - IDP functions for deep packet security
- Wire speed head ends to protect VoD streamers
- Application accelerators for SSL Offload and HTTP
load balancing of VoD navigation portal - Enhanced security
- OpEx reduction via lower remediation costs
- Enhanced revenue through reduced churn
Multicast Encoder/ Streamer
VoD RTSP Control
VoD Navigation
VoD Streamers
SSL HTTP Assist
Middleware
Head-End Router
Middleware
Firewall
Multicast Access control
DHCP
Head-End (Broadcast and Routing)
Content Management Platforms
21Cost of a Security Incident
- Calculations for estimating the costs associated
with a security incident range from simple to
complex
22Economics of Architecture Summary
- Necessity of aggregation is dependant on the
existing architecture and DSLAM utilization - Architectural decision for Centralized verse
De-centralized approach are dependent on many
factors which must be considered together and
over the life of the project - Important not to overlook
- Zero touch provisioning and online customer care
- Adding new services quickly to move customers to
higher margin services - Security and protection of the IP infrastructure
23Thanks!Questions?