Professional Documents
Culture Documents
Comments of
1
TNC 2011-77, paragraph 12, sub (i).
2
Commission letter Re: TNC 2011-77, Requests to modify the scope and terms of the proceeding, March
11, 2011, page 2.
3
Cable Carriers’ comments, February 8, 2010, Telecom Notice of Consultation 2009-261, paragraphs 35-
40.
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page 2 of 11
• Section 2.0 explains how an economic ITMP such as UBB fulfills the
Commission’s consumer fairness principle;
14. The Commission’s first principle of fairness for consumers dictates that consumers
who have lower volume use and place fewer demands on the available network
capacity should not fund the costs of providing sufficient bandwidth to meet the
demands of higher volume users. Economic ITMPs satisfy this principle.
2.1 Regulatory background
15. In Telecom Regulatory Policy 2009-657 (TRP 2009-657), the Commission
recognized that Canadian carriers that provide high speed internet access service
over their networks have a legitimate right and a responsibility to manage these
networks. Network management needs to balance the demands of users and the
ability of the carriers to fulfill that demand. In particular, the network needs to be
managed to minimize the risk of network congestion and the adverse effects it can
have on the quality of service.
16. The Commission established its policy framework for ITMPs in TRP 2009-657. The
Commission agreed that the networks of the facilities-based providers experience
congestion and recommended that the companies first seek to invest sufficiently to
minimize congestion. However, the Commission recognized that “investment alone
does not obviate the need for certain ITMPs”.4
17. TRP 2009-657 described two main types of ITMPs – technical and economic.
Technical ITMPs include slowing down a user’s traffic, prioritizing traffic,
and detecting heavy users in order to limit their bandwidth.
4
TRP 2009-657, para. 36.
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page 4 of 11
18. The Commission noted in TRP 2009-657 that some types of technical ITMPs could
give rise to concerns respecting discrimination or preference, while “economic
ITMPs would generally not be considered unjustly discriminatory”.6 The Commission
was supportive of economic ITMPs because they “provide greater transparency” and
“match consumer usage with willingness to pay”.7
19. ITMPs of various forms are recognized among regulatory authorities internationally
as measures that minimize the risk of congestion. See Appendix A for excerpts from
recent statements by regulators in the United States, United Kingdom, France and
Europe.
2.2 Economic ITMPs satisfy the Commission’s consumer fairness principle
20. Economic ITMPs provide price signals to end-users that promote an appropriate
allocation of the limited capacity available on the network. The price signals permit
consumers to purchase a level of service that matches their preferences.
21. The current market allows a consumer who places a high value on accessing
applications and content requiring significant bandwidth to select an appropriate
service level and associated usage allotment. Conversely, a consumer with a
modest usage profile can select a service at a lower price point that corresponds to
the lower value assigned to accessing such applications and content. As a result,
lower volume users do not fund the higher bandwidth requirements of high volume
users.
22. The fact that economic ITMPs satisfy the Commission’s first principle of consumer
fairness can be further illustrated by considering two simple pricing schemes: flat
rate and usage based.
23. A service that is priced at a flat rate is efficient for both the supplier and consumer
when the cost to provide the service is largely fixed and demand is homogeneous.
Flat rate pricing creates incentives for consumers to increase use even if they derive
little or no additional benefit since they incur no incremental cost associated with that
use. The service provider will be able to profitably supply the level of demand
generated under a flat rate plan since the increased demand does not increase its
5
TRP 2009-657, footnotes 5 and 6.
6
TRP 2009-657, para. 40.
7
Ibid.
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page 5 of 11
costs. This model is more likely to apply when consumer demand can readily be
satisfied at the existing level of supply without decreases in quality.
24. If the costs to provide the service vary with the level of demand, and demand levels
are not homogeneous, then usage based pricing is more efficient. Usage based
pricing signals to consumers they should use only the amount of the service that
they need. A consumer who chooses to increase use should derive an increase in
value. Consumers who derive greater value from the increased use will be allocated
more of the service or product from the available supply in return for which they will
be willing to pay more. Consumers who want to use less will enjoy savings, relative
to a flat rate pricing model. In economic terms, the price signals an efficient
allocation of the limited resource. The supplier will have an incentive to increase
supply to serve the growth in demand generated by a group of consumers since the
growth in demand generates revenues that recover the additional costs of increasing
supply.
2.3 Economic ITMPs applied to high-speed access services
25. Capacity available for internet service is a finite resource at any given time. It is
entirely plausible for all capacity to be consumed. If some consumers are using a
large volume of capacity, for example, by uploading or seeding multiple peer to peer
(P2P) files, then this capacity is not available for other consumers at that time. Time-
sensitive applications will be degraded as a result.
26. The costs of providing internet service consist of large fixed costs in the network and
costs that are sensitive to changes in demand. As long as demand remains at or
below available capacity, costs will remain largely fixed. Cable carriers continuously
monitor the capacity used on their networks. When it appears that demand is likely
to exceed the available capacity, then additional investment is required.
27. Cable Carriers augment network capacity primarily by splitting nodes. This reduces
the number of subscribers that share the available capacity on the last mile portion
of the network. Fibre facilities must also be extended to serve the additional nodes.
28. The Cable Carriers have filed extensive costing information as part of the
proceeding Follow-up to Telecom Regulatory Policy CRTC 2010-632 (Follow-up to
TRP 2010-632) demonstrating that the costs of providing TPIA, and high-speed
access more generally, are sensitive to changes in traffic volumes.
29. The provisioning of internet can be contrasted with provisioning traditional linear
cable television channels where all channels are broadcast to subscribers
simultaneously. The costs of distributing television channels broadcast over the
cable network are driven by the number of channels supplied, not the number of
consumers who are watching those channels at any given time.
30. The market for high speed internet services is also characterized by a broad range
of consumer usage profiles and growth in demand across most of these profiles.
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page 6 of 11
with lower usage patterns derive less value from the service and would be less likely
to subscribe at the higher price. Higher volume consumers would likely continue to
subscribe and might further increase their usage since they would have an economic
incentive to do so in order to gain further value from the service. Network
congestion might still occur if the remaining subscribers increase usage sufficient to
offset the reductions of departing subscribers. The service provider would not avoid
the costs of network congestion, and would experience a net reduction in revenues
due to the lost subscribers. Moreover, the lower volume consumers who dropped
the service would be worse off. This would not be consistent with the Commission’s
consumer fairness principle.
37. One method to avoid network congestion with flat rate pricing is to set a limit on the
amount of usage permitted at that price. The service provider can employ technical
ITMPs to curtail use beyond that limit. Consumers who exceed the limit experience
lower service quality (e.g., speed, bandwidth). Because the service is flat rated,
these consumers may not have the option to pay an additional fee to restore their
service. This approach could limit the ability of consumers to have full access to the
internet.
38. The Cable Carriers have applied technical ITMPs only with respect to upstream
traffic. These measures are applied primarily to control the extent to which peer to
peer applications consume the limited capacity available on the upstream
bandwidth.13 Technical ITMPs that limit capacity for P2P upstream applications
improve the quality of service of time-sensitive applications that require upstream
bandwidth, such as VoIP and online gaming.
39. Network congestion can also be controlled by applying usage based pricing.
Applying a usage based component to the price can supplement or take the place of
a technical ITMP. Consumers must choose whether to incur additional costs in
return for using more capacity. Higher volume consumers may choose to pay more
if they want to increase their usage.
40. Usage based pricing permits consumers to receive a consistent level of service
without interference regardless of their usage level. By comparison, a technical
ITMP would interfere with the service quality if the consumer’s use exceeded the
limit.
41. A common form of usage based billing as an economic ITMP is two part pricing.
The first part is flat rated and includes a fixed amount of usage. The second part is
usage based pricing that applies to use that exceeds the amount included at the flat
rate price. This approach provides consumers with the benefit of predictable billing in
return for self-managing their usage within the limit. It also allows usage over that
limit in return for additional payment.
13
Further details on technical ITMPs are provided in the responses to Cogeco(CRTC)4Feb11-107,
QMI(CRTC)4Feb11-106, and Rogers(CRTC)4Feb11-106, filed in the Follow-Up to TRP 2010-632.
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page 8 of 11
42. In order to be effective as an economic ITMP, the usage based price component
needs to be established so as to discourage use above the set limit. The price
should incent use in excess of the limit only to the extent that the consumer would
gain significant value from that usage. If the price is set substantially below the
consumer’s value, it will have little influence on usage. It follows that the price does
not necessarily reflect the cost of supplying the network capacity.14
43. The Cable Carriers have applied economic ITMPs in the form of limits on usage for a
number of years. In some cases, specific charges for usage in excess of the cap
have been applied as well. The experience with UBB has been that the vast
majority of consumers do not exceed the cap applied to their associated service tier.
Among the small percentage of consumers with higher volume usage, some choose
to moderate their use while others switch to a higher tier with higher usage quota to
accommodate their usage profile.
44. It has been suggested in some forums that usage-based pricing should be targeted
either at certain time periods when traffic volumes are at their peak or in specific
locations where the volume of traffic is highest. Such a de-averaged pricing
approach is not equitable and may not be as effective in influencing consumer
usage.
45. Peak periods can span broad time frames. Information filed in Follow-up to TRP
2010-632 indicated that the peak period can last several hours, starting in the late
afternoon and continuing after midnight.15 See also the statistics compiled by
Sandvine provided in Appendix B, Chart B3. Consumers may not have the flexibility
in their daily schedule to shift their usage away from the peak period. Higher prices
in peak periods would penalize consumers that were not able to shift their usage to
off-peak periods. There would also be additional significant costs to implement peak
period pricing.
46. Usage-based pricing can be effective at moderating congestion in peak periods
without necessarily targeting peak periods. Consumers have been found to
moderate their use in peak periods in response to non-time sensitive pricing
incentives.
47. Consumers unlikely to manage their usage in response to pricing plans that targeted
specific geographic locations where the volume of traffic is higher than average, at
least in the context of wireline high-speed access services. For example, high
volume consumers are not likely to switch residences to a lower volume or less
congested neighbourhood. Setting higher prices for services in neighbourhoods
characterized by high volume usage would mean consumers in those areas would
14
This relationship is discussed further in the Cable Carriers’ comments filed in response to Telecom
Notice of Consultation 2010-803, November 29, 2010, para. 16.
15
See for example the response to Shaw(CRTC)4Feb11-102 TRP 2010-632.
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page 9 of 11
be charged higher prices that may not reflect their own usage profiles. Rather, these
consumers would end up paying more because of the usage profiles of their
neighbours.
16
Telecom Order CRTC 2000-789, para. 8; Telecom Decision CRTC 2006-77, para. 249.
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page 10 of 11
network capacity. Absent a change in the TPIA tariffs, there is no means to recover
the increase in costs caused by the higher usage of the Small ISPs’ end-users.
53. This demonstrates why economic ITMPs need to be applied in a similar manner to
the retail and wholesale services where both services are provisioned using the
same network.
54. Videotron has applied UBB to both its retail and wholesale high-speed access
services for a number of years. Videotron described the implications of moving
away from this symmetrical treatment in a recent interrogatory response.
It is important to note that Videotron’s TPIA cost model assumes that retail and
wholesale end-users have precisely the same usage profile per speed tier.
Should a policy change result in wholesale end-users being subject to more
relaxed UBB thresholds or lower UBB rates relative to retail end-users, then we
would no longer expect wholesale end-users to have the same usage profile as
retail end-users. Wholesale end-users would be expected to consume more, and
Videotron’s monthly TPIA rates would have to be adjusted in consequence.17
17
QMI(CRTC)4Feb11-106, (e), filed in the follow-up proceeding to TRP 2010-632.
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page 11 of 11
number of higher volume consumers, then it is more likely to exceed the average
usage limits and incur additional usage charges under the TPIA tariff. Additional
chargers should apply when large volumes of traffic are generated by the end-users
of a Small ISP. Otherwise, usage based billing would not be an effective economic
ITMP.
59. The Cable Carriers are of the view that Small ISPs would have an appropriate
degree of flexibility under an aggregated model. The threshold at which charges
would apply to a Small ISP would depend on the average volume of use for all the
end-users it served per tier, rather than the usage of each individual end-user.
Additional usage charges would be based on the aggregated usage that a Small
ISP’s end-users generated in excess of the threshold per tier.
60. The Small ISPs would be able to offer competitive alternatives with prices, usage
quotas and/or charges for additional usage that differ from the Cable Carriers’ own
retail services. This would be consistent with the second principle enunciated in
TNC 2011-77. At the same time, the Small ISPs would have an economic incentive
to establish appropriate measures that moderated the overall demand generated by
their end-users.
61. Pricing models that would apply usage-based billing to TPIA by time of day would
provide less flexibility for competitive alternatives. In addition, time of day pricing
would affect some consumers who may be unable to shift their usage to off-peak
periods, as noted in Section 2.3. However, an aggregate model applied to TPIA
without targeting peak period usage would not preclude the Small ISPs from offering
retail services with usage and pricing elements that are targeted at controlling use in
peak periods.
62. The Cable Carriers are not in a position to propose specific pricing plans for usage
under an aggregated model at this time. However, as a general proposition, the
wholesale pricing should reflect retail prices for UBB and not be based on a narrow
incremental Phase II cost methodology. The reasons for this were provided in the
Cable Carriers’ submissions filed in Telecom Notice of Consultation 2010-803.18
63. The Cable Carriers remain of the view that setting wholesale UBB prices based on a
narrow assessment of incremental costs would be detrimental to ensuring that UBB
functions as an effective economic ITMP on the shared resources of the cable
network. The integrity of economic ITMPs should not be sacrificed simply to
artificially promote resale competition.
18
Cable Carriers’ comments TNC 2010-803, November 29, 2010, paragraphs 12 to 16; and December 9,
2010, paragraph 23.
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page A1 of A2
APPENDIX A
19
FCC, In the Matter of Preserving the Open Internet Broadband Industry Practices, Report and Order,
December 21, 2010, para. 72; available at
http://www.fcc.gov/Daily_Releases/Daily_Business/2010/db1223/FCC-10-201A1.pdf
20
Ofcom, Traffic Management and ‘net neutrality’, A Discussion Document, June 24, 2010, para. 4.11,
4.12; available at: http://stakeholders.ofcom.org.uk/consultations/net-neutrality/
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page A2 of A2
APPENDIX A
that, when an end user exceeds this “reasonable” consumption threshold, it could
result in him having his access speed reduced, or he may be billed an overage
charge on top of his flat fee. This type of practice already exists, particularly on
mobile and cable networks, but the way it is applied is generally unclear. It
seems neither opportune nor relevant to forbid operators from engaging in this
type of practice. To the extent that it appears that 5% or 10% of end users
consume more than half the bandwidth on electronic communications networks,
it may be preferable in certain cases for ISPs to implement this type of system to
ensure that the behaviour of a minority is not detrimental to the quality of service
provided to the majority of end users.21
Body of European Regulators for Electronic Communications (BEREC)
The flat rate subscriptions currently provided may in case of high data volume
usage at peak hours not cover all the costs of the ISP. The solutions commonly
envisaged by operators are either to limit the data volume or to increase the
price. Limiting the data volume or throughput rate may either be independent of
data type (data volume caps or bandwidth limits) or dependent of data type (e.g.
throttling of p2p file sharing). The first method does not constitute a departure
from the principle of net neutrality; while the second method presumably does as
specific data types receive a different treatment than other traffic. There are
many existing models to address an imbalance between price and cost according
to the first approach – for instance offering price-differentiated access
throughputs (this is generally well accepted and raises no net neutrality
concerns). Another concrete illustration of the first method is that some ISPs
include a “Fair Usage Policy” clause in their end user contracts, which is often
expressed in very general terms but is used to justify the application of the
mentioned mechanisms. Operators generally state that fair use policies are
intended at preventing some “extreme” profiles of consumption (of a minority of
users) to degrade the Internet experience of others.22
21
ARCEP, Internet and network neutrality, Proposals and recommendations, September 2010, page 33;
available at: http://www.arcep.fr/uploads/tx_gspublication/net-neutralite-orientations-sept2010-eng.pdf
22
BEREC, BEREC Response to the European Commission’s consultation on the open Internet and net
neutrality in Europe, September 30, 2010, page 15; available at http://erg.eu.int/doc/berec/bor_10_42.pdf
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page B1 of B4
APPENDIX B
NETWORK USAGE
Network capacity is a finite resource, at least in the short term. Investment in the
network, such as splitting nodes in the cable plant, can increase capacity over longer
time frames. However, the Cable Carriers’ ongoing efforts to increase capacity are
being met by increasing traffic volumes that are filling the available capacity. The
following statistics demonstrate that growth is expected to continue.
Chart B1, based on Cisco’s VNI data, illustrates the expected growth in usage in
Canada, broken down by traffic type.23
CHART B1
1000
880.7
800
707.1
s 676.2
te
y 553.3
b 600 546.9
a
t
e
P 427.5
375.3
400 322.4
256.3 169.1
185.2 139.6
200 103.4 115.5
95.8
79.6
66.4 207.0
156.2 184.5
105.2 128.5
82.2
0
2009 2010 2011 2012 2013 2014
23
Cisco VNI Forecast Widget, analysed for Canadian consumer internet traffic; see
http://ciscovni.com/vni_forecast/advanced.html
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page B2 of B4
APPENDIX B
NETWORK USAGE
The growth in overall consumer traffic volumes in Canada between 2008 and 2010 are
on par with those estimated for the United States and Japan and will exceed those in
the United Kingdom, France, Germany and South Korea, according to Cisco VNI
statistics. The relative growth rates for Canada and other countries are shown in the
following chart.
CHART B224
Statistics from Sandvine, Fall 2010 Global Internet Phenomena Report provide further
insights as to the distribution of internet traffic on the North American fixed access
connections (as opposed to mobile), both by time of day and by user profile.25
Sandvine’s statistics indicate that network capacity consumption is spread across a
relatively broad time period, with the vast majority of subscribers remaining online
throughout the day. These factors combine to place considerable pressure on available
network capacity.
24
Source: Lemay-Yates Associates Inc., The Performance of Canada’s Consumer Broadband Networks
in 2010, July 13, 2010; Report presented to Rogers Communications Inc., available at
http://www.lya.com/en/spotlight/documents/FINALLYARogersNetworkPerformanceReportJuly132010.pdf
25
Sandvinde, Fall 2010 Global Internet Phenomena Report, available at:
http://www.sandvine.com/news/global_broadband_trends.asp
Review of billing practices for wholesale
residential high-speed access services
March 28, 2011
Page B3 of B4
APPENDIX B
NETWORK USAGE
The following chart indicates that there is only a brief period during the day when traffic
volumes are relatively low (blue line in the chart). By midday, traffic volumes have
reached 70% of the maximum and exceed 90% by early evening. It is also worth noting
that at least 75% of subscribers remain connected even during the off-peak period (red
line in the chart).
CHART B3
Internet Use by Time of Day - North America 2010
APPENDIX B
NETWORK USAGE
volume of traffic in a month. At the other end of the usage profile, the 50% of end-users
with the lowest volume account for only 5% of the total monthly traffic.
The following chart provides a further breakdown by upstream and downstream traffic
use. As this chart indicates, the variation in usage profiles is skewed even more heavily
for upstream, with the top 5% accounting for 70% of upstream traffic.
CHART B4
Upstream Downstream
Source: Sandvine, 2010