Explanation
of Potential Unbundling Requirements on SBC’s Project Pronto Network
Architecture.
One of the primary concerns regarding SBC’s Project
Pronto deployment is the impact on the existing, and planned network
infrastructure. Unbundling requests before
the FCC and various State Commissions are focused primarily upon one simple
issue: the establishment of a mandate that SBC provide CLECs use of the Project
Pronto network architecture to support any service that CLECs so desire. This one issue is the driver behind all CLEC
requests, including such issues as line card “collocation” and the piece-part
unbundling of Project Pronto. For
example, CLECs are attempting to establish the capability to place any line card
of their choice in order to offer differentiated services from SBC’s ADSL
service that the Project Pronto architecture was not designed to efficiently
support.
There are very real consequences of taking such
action. First, any requirement that
allows CLECs to place their own line cards in ILEC equipment and/or mandates
that SBC provide any service, feature or function of the Project Pronto network
architecture would add substantial costs to Project Pronto deployment. In fact, such costs are estimated to be so
significant as to make the ADSL service that is currently offered by SBC for
all carriers (CLECs included) uneconomic in contrast to competitive services
such as cable modems. Therefore, SBC
would have no incentive to move forward with ongoing Project Pronto
deployment. Second, in addition to the
costs of supporting these alternatives, CLEC requests would serve to limit the
availability of SBC’s ADSL service over the architecture and/or serve to
detrimentally impact the quality of service that could be provided over the
Project Pronto infrastructure.
The additional costs that would be created by the CLECs requests are driven by two basic factors: (i) stranded capacity created by CLEC line card collocation, and (ii) additional capacity that would be required to augment the network to support the CLEC’s desired services.
Allowing CLECs to “collocate” line cards in order to
support differentiated services impacts the network in two different
manners. First, line card collocation
results in stranded capacity in the Project Pronto remote terminal (RT)
locations. Consider the simple example
of one CLEC collocating one line card to offer a differentiated service. In the Project Pronto equipment, one line
card is capable of supporting four end-user customer services. However, if a CLEC were to collocate a line
card in order to offer one end-user a differentiated service from SBC’s ADSL
service, that line card may not support the intended base of 4 end-user customers
or such line card (where it to be capable of supporting 4 end-user customers)
may only be serving one particular end-user.
The end result is stranding of additional capacity that otherwise would
be available in a particular slot in the RT that could have been used to
provision additional ADSL service.
This is a direct opportunity cost from SBC’s
perspective. Whereas one slot in one RT
previously could have supported four end-users, now this one slot serves
potentially one end-user (unless and until the CLEC adds an additional customer). Given that each RT is limited to a finite
number of slots, the result is stranded capacity and a potential reduction in
the number of end-users that could be served by the RT slot. Conversely, in order to maintain service
availability to the intended customer base, SBC would incur additional costs to
provide service, as additional card slots (e.g.,
additional RTs, fiber and associated electronics) would have to be placed into
the network.
This effect is magnified by the fact that not every channel bank in a given Project Pronto RT site is wired out to each and every Serving Area Interface (SAI) location subtending that RT site. As shown in Attachment 2 which illustrates the Project Pronto network architecture, the ADSL channel banks in a Project Pronto RT site are wired out to an SAI location, which then serves several end-user customer premises. The diagram in Attachment 2 is a representative example. However, the Project Pronto RT sites typically serve three to five SAI locations. Therefore, in some instances, it may be necessary for a CLEC to place cards in multiple channel banks to reach customers served by differing SAIs. The end result is an additional loss in overall ability to serve end-user customers.
Furthermore, this loss of end-user capacity would
not only impact SBC’s ability to provide service, but also other carriers. This is due to the fact that SBC already
provides all CLECs non-discriminatory access to its Broadband Service product
offering that is used by its affiliates to provision ADSL service over Project
Pronto. Therefore, the simple act of
one CLEC reserving an entire slots worth of capacity for its “differentiated”
service offering reduces the amount of available capacity for any other
carrier, SBC and other CLECs alike.
The only means to mitigate this lost capacity would
be for SBC to augment its network with additional capacity. However, these costs are so significant as
to render the continued deployment of Project Pronto uneconomic. For example, in Illinois, where this
requirement was first imposed on SBC,[1]
the estimated capital cost due to stranded capacity created by CLEC line card
collocation would be in excess of $50 million.
This cost, when spread across the projected demand for ADSL across the
state of Illinois would drive ADSL service over the Project Pronto network
architecture out of the market.
Furthermore, this cost does not even contemplate the potential cost of
system modifications to support CLEC line card “collocation”. In fact, the estimated additional capital
costs to offset stranded capacity created simply by CLEC line card collocation,
spread across the expected ADSL take rates over Project Pronto for all carriers
in Illinois, could potentially more than double the cost for the Broadband
Service product (offered on a parity basis to all carriers). In order to understand the order of
magnitude of this increase, one must also understand the additional costs
associated with providing an end-to-end ADSL service. When one adds SBC’s costs to reimburse the ISP and the necessary
additional overhead, marketing and administrative costs incurred by SBC’s
Advanced Services Affiliates, the cumulative price tag to the end-user customer
for the ADSL service as provisioned over Project Pronto would increase to more
than $80 per month. Given that the
going market rate for broadband Internet access generally across the country is
$50 or less per month, it is easy to see why SBC viewed the continued
deployment of Project Pronto in Illinois as uneconomic in light of this
requirement.
In addition to the costs created by CLEC line card
collocation due to stranded capacity, there is the related issue of the costs
created by the services provisioned using these CLEC placed line cards (e.g.
the differentiated services desired by the CLECs). For example, there is a limited amount of bandwidth from each RT
site to the serving wire center. The
provision of a non-ADSL service (as requested by a CLEC) would allocate this
finite amount of bandwidth in different manners. In some instances, CLEC desired services would effectively lead
to the premature exhaust of this bandwidth and thus lead to increased costs.
Consider that the amount of useable bandwidth from a
given RT site to the central office is typically limited to 135 Mbps of
available bandwidth. Now consider the
situation in which a CLEC wanted to deploy an SDSL service in a given RT site,
were such a capability made available by SBC’s vendors in the future. One use of an SDSL service could be to
provide the equivalent to a DS1 (1.544 Mbps) service to a business for the
purposes of data transmission between two points. A typical business class SDSL solution generally requires a
Constant Bit Rate (CBR) ATM quality of service. This is due to the fact that a business using SDSL would
typically not want to share bandwidth with other end-users. In most instances, a business would instead
require a guaranteed speed between two points.
This is in direct contrast to consumer high-speed Internet access, which
typically is allocated bandwidth with an Unspecified Bit Rate (UBR) quality of
service.
The difference between UBR and CBR is simple. With a CBR service an end-user is dedicated
(guaranteed) at all times a fixed, constant amount of bandwidth. In contrast, a UBR customer is only provided
the available amount of bandwidth when they access the Internet. For example, a business, in order to
transport large amounts of data on a real time basis, may need a constant,
dedicated amount of bandwidth. However,
a consumer, typically does not need a dedicated amount of bandwidth. This is because when a consumer accesses the
Internet, typically they will access a website, download content from that
website, read that content and then move on to another location on the
Internet. During this time period in
which the consumer is reading the content on a particular website/webpage, the
consumer is generally not downloading/uploading any information. Therefore, a typical consumer “web surfing”
from their residence would have no need to be allocated a constant portion of
the available bandwidth. For this
application, an unspecified allocation of bandwidth is the more appropriate
level of service.
The fundamental design and economics of SBC’s
Project Pronto deployment are predicated upon the offering of consumer,
high-speed Internet access (and as such a UBR type of offering). As explained above, with a UBR offering, an
end-user ADSL service does not occupy a dedicated portion of the available
bandwidth within the Pronto network architecture. Therefore, SBC can offer many more end-users the equivalent to a
high speed DSL service than would theoretically be possible using a CBR
application.
The reason for this is straightforward. For each customer provided a CBR service,
the service utilizes a portion of the available bandwidth that could otherwise
have been used to provision high-speed ADSL service for consumer Internet
access. Consider a situation in which a
carrier was offering a 1.544 Mbps downstream service to an end-user. If this 1.544 Mbps of bandwidth were
allocated in a dedicated manner to a given end-user (as a CBR service would
require) simple math would indicate that the most number of these high bandwidth
services that could be provisioned would be 135 Mbps (the finite amount of
bandwidth provisioned by SBC at each RT site) divided by 1.544 Mbps or
approximately 85 end-user services. The
simple fact is that for each CBR service provisioned, at whatever speed, there
is a reduction in the available capacity for UBR services provisioned over the
architecture.
In contrast, a UBR service allows SBC to “oversubscribe” its network. Over-subscription takes advantage of the fact that generally end-users accessing the Internet are not downloading all of the time. Therefore, while one end-user may be online and downloading, another end-user may not be. The end result is that many more end-users can be provisioned service over this architecture than would otherwise be possible using a CBR type of application.
For example, as illustrated in the chart below,
approximately 672 end-users are capable of ADSL service in a given NGDLC RT
site[2]. With a UBR class of service, it is feasible
to transport all 672 of those end-users ADSL service over the 135 Mbps of
available bandwidth and still offer a robust, high bandwidth downstream service
to a residential end-user (taking advantage of over-subscription). Therefore, via over-subscription, SBC can in
many instances offer high speed Internet access to upwards of 672 or more
end-users from one RT site. This in
contrast to the fact that were a CLEC to provision 85 end-users the
aforementioned 1.544 Mbps CBR service, the CLEC would in essence exhaust the
available bandwidth at a given RT site, leaving no capability for SBC to
provide a standard internet access service to any end-users – and thus further
requiring the augmentation of its network.
Graph: The following graph illustrates how the shift to provide higher
level CBR services at certain volumes dramatically impacts the number of
end-users that can be provided UBR service.
A simple analogy that may help explain the manner in
which the use of differentiated services, such as CBR, impacts SBC’s Project
Pronto network is a four-lane highway.
In this analogy, all of the cars traversing that highway freely share in
all of the four lanes on that highway.
Now consider if one of the lanes on this highway was dedicated to one
car due to a mechanical failure. As we
all know, this leads to a traffic jam. This is no different than what could
occur in a telecommunications packet network.
The cards themselves could be considered individual packets representing
an end-user customer. If one of the
lanes in the packet network were allocated to this one end-user – one car – via
the provision of a dedicated CBR service, all of the other cars in this network
are forced to occupy a smaller highway in essence. The end result is that these end-users receive a lower quality
service than could otherwise be obtained.
The only means to compensate for this slower service would be to
increase the size of the highway – in essence to add additional bandwidth to
the packet network – which just as the cost of growing the size of the
interstate highway system is cost prohibitive, the same would apply to
incrementally growing SBC’s Project Pronto network architecture.
In fact, one of the precise economic principles upon
which Project Pronto was based was the ability of SBC to over-subscribe its
network. Lacking this ability, the end
result would be less availability to end-user customers. Thus, the capital cost of the Project Pronto
deployment would be spread across fewer end-users than had been planned. This would directly lead to higher rates
for the ADSL service as provided over the Project Pronto network
architecture. The result of such an
increase in price could potentially impact the competitive viability of ADSL
service in contrast to other services such as cable modems. This could make the Project Pronto deployment
uneconomic as higher costs and indirectly higher prices for service could
potentially lessen demand and thus damage SBC’s ability to recover the costs of
its investment.
For example, as has been addressed above, the
provision of higher bandwidth services desired by CLECs (including high
bandwidth CBR service), presumably to serve business customers, would lead to
the allocation of bandwidth in a manner that would lead to a reduction in the
number of end-users that could theoretically be served by a given RT. Also, as stated above, the only means to
rectify this problem would be for SBC to augment its network by deploying
additional bandwidth. This would mean
additional capital expense in the form of fiber and associated
electronics. The cumulative price tag
of the CLECs requests is staggering.
This cost when added to the aforementioned costs of CLEC line card
collocation would more than double the cumulative costs of providing ADSL
service over the Project Pronto network architecture. This in a price sensitive
market currently pricing services at or less than $50 per month. It is relatively easy to see why continuing
deployment in Illinois under this scenario would make little or no economic
sense from SBC’s perspective.
Furthermore, the provision of more robust, bandwidth eating services over this architecture, in lieu of augmenting the network with additional high cost bandwidth capacity, would only serve to detrimentally impact the overall availability and quality of service to consumers using UBR service for the purpose of internet access. Consider the same highway analogy as discussed above. If the same number of cars is forced onto a highway with one less lane the end result is that each car experiences a backup. In a telecommunications network this equates to a slower-speed service. There are two means to rectify this problem: to grow the capacity of the network, which as was mentioned above is cost prohibitive, or to simply provide less cars (end-users) access to the highway (network). If one is not willing to take these steps then each end-user will experience degradation in their service performance.
The principal objective of CLECs requests in seeking access to and use of SBC broadband network is to pursue their business models. This would be nothing more than a usurpation of SBC’s investment. In fact, virtually all of the CLECs desired services in this space are designed to focus upon the business customer. CLECs are proposing to shift SBC’s investment from its intended base of the mass market (consumers and small businesses) to focus upon business customers providing higher margins. If this result were to occur on a widespread basis this would simply increase options for businesses (which already enjoy a highly competitive telecommunications market place) to the detriment of consumers. Use of the Project Pronto investment in the manner proposed by CLECs would in essence be the use of the architecture for a purpose other than that originally planned for by SBC, and would fundamentally alter the business case economies underlying SBC’s plan Project Pronto investment.
[1] The Illinois Commerce
Commission has since changed its position to not require CLEC line card
“collocation” over Project Pronto based upon the results of a rehearing of its
original decision in the line sharing tariff proceeding.
[2] As illustrated in
Attachment 2 there are approximately 3 channel banks capable of providing ADSL
service in a given Project Pronto RT site.
Each channel bank contains 56 slots.
Using the forward looking ADSL card that provides four ports, capable of
serving four customers, per card – the total physical capacity of a given RT
site in a typical configuration is 56 x 3 x 4 or 672 total ADSL customers in
most instances. This is the current
configuration supported by SBC.