Response to NBN Co consultation paper: proposed wholesale fibre bitstream products #1

This is merely a high-level summary of my formal response to NBN Co’s first industry consultation (warning video stream) as part of their industry briefing sessions. This first open engagement is centered around their “proposed wholesale fibre bitstream products”. This summary and my response follow the same format and section numbering as the original consultation paper. Unsurprisingly, more detail can be found in my actual submitted response. For each section, I’ll provide a summary of the material as documented by NBN Co, followed by a summary of my response to the corresponding section and any questions posed within.

Section 1: Introduction
NBN Co provided some high-level background behind their mandate to build a next generation wholesale broadband access network. They also provided an introduction to the process of openly engaging industry as part of their network design and defined the scope of this particular engagement. Finally a top-level overview of their proposed bitstream services, a local Ethernet bitstream product (LEB) and an aggregated Ethernet bitstream product (AEB) was provided.

My response basically acknowledged the process of engagement and the scope of this engagement. Although discussions regarding ‘pricing’ were placed out-of-scope by NBN Co, my response argued that some analysis of pricing models is necessary in order to properly evaluate some architectural decisions. I also noted that wholesale product models need to carefully balance the principles of supposedly “perfectly simplifying” (aka Occam’s razor extended to K.I.S.S.) against the principle of “maintaining necessary complexity” (aka Einstein’s razor or don’t over-simplify). A good litmus test for network solutions (and others as well) is to ensure that the simplest environments can always be solved in the simplest manner, whilst simultaneously ensuring that most demanding and complex environments can still be accommodated in a reasonable fashion.

Section 2: Building a fibre access network
NBN Co introduced a reference diagram to illustrate the various components involved in constructing a fibre serving area (FSA). These included identifying different residential dwellings (free standing houses and multi-dwelling units), different fibre deployment approaches (aerial and ducted), different fibre segments (drop fibre, local fibre and distribution fibre) as well as the active equipment plant (ONT, OLT, Fibre Distribution Hub, Fibre Access Node and RSP Backhaul). This diagram also included support for delivery via Cu-based VDSL within some multi-dwelling units. NBN Co identified a number of definitions and issues that will need to be addressed as part of future engagements. My response basically agreed with their list, noting that many of these ‘yet-to-be-determined’ parameters will need to be defined and communicated at some stage–the earlier, the better of course.

Section 3: NBN Co’s overall product objectives
NBN Co listed the key criteria and objectives that guided their product design (and in fact much of the network design and business operations I suspect). These fundamental objectives are:

• Coverage – aim is to deliver a consistent service to retail service providers (RSPs) that ensures a consistent end-user experience throughout Australia.
• A level competitive playing field – aim is to support healthy retail competition, allow differentiation and innovation by RSPs, to support secure simultaneous delivery of multiple applications and to promote maximum end-user choice.
• Cost effectiveness – aim is to limit NBN Co functionality to only those parts that are critical, to focus on uncontested infrastructure only and to align with global standards and proven technology.
• Customer care for RSP’s and end-users – aim is to ensure a high quality, reliable, available and secure network with appropriate support for critical and emergency services as required.

My response basically agreed with these stated objectives.

Section 4: Choice of layer in the vertical technology stack
This section analyzed the advantages and disadvantages associated with the various approaches for offering wholesale products within a broadband access network. Each of the options (layer-1 (dark-fibre) vs layer-2 (Ethernet) vs layer-3 (IP)) were weighed up against various attributes from the objectives section (coverage, support for differentiation and innovation, support for healthy retail competition, support for maximum end-user choice, support for access to multiple applications, simplicity, focus on uncontested infrastructure and finally reliability, availability and security). Questions were raised around the final choice to focus on a layer-2 wholesale delivery model as the preferred compromise.

My response to this section generally agreed with the choice of a layer-2 wholesale service under the assumption that a vibrant and competitive market for wholesale layer-3 evolves independently (something that recent comments from the engagement sessions seem to support). I took the opportunity in my response to this section to introduce the concept of local community service providers as an important scenario for NBN market analysis. Basically I believe that local community services such as health, information, council, libraries, schools, volunteer organisations and small businesses (such as independent video rental stores, news agencies, community radio/video/education) should be actively encouraged by NBN Co to connect with their local communities rather than discriminated against commercially by the larger players. In my opinion, this is a very important sanity check against whatever product delivery options are chosen. I refer back to this local community scenario several times in my response. It’s important to note however, that the same conditions equally apply for a national retail/wholesale service provider who is also seeking to offer some form of highly localised service capability, such as content caching for instance.

Section 5: High level technology standards
NBN Co confirm that the planned network build will be based on a GPON (Gigabit Passive Optical Network) with MEF (Metro Ethernet Forum) based Ethernet services. Questions were raised around alternatives such as GEPON (Gigabit Ethernet PON) and requirements/triggers for offering point-to-point fibre products.

My response concurred that GPON and Ethernet are world best practice for FttH (Fibre to the Home) access networks and listed a number of key criteria to consider around the Ethernet service description (framing and tagging, priority, frame-size and OAM&P (Operations, Administration, Maintenance and Provisioning) practices). I also agreed that some allowance for point-to-point services be made within the passive fibre plant and listed the following, at least foreseen, triggers for offering point-to-point solutions:

• Specialised security requirements, including but not limited to quantum encrypted links;
• Specialised protocol environments, including but not limited to fibre-channel based data silo wharehouses; and
• Specialised bandwidth environments, where the dedicated bandwidth requirements exceed the levels typically found in the average mass-market (currently defined as 100Mbps minimum).

Section 6: Location of Points of Interconnect for NBN Co wholesale fibre network
NBN Co re-used a very good abstract network architecture diagram from the Communications Alliance as a reference model for identifying the logical interconnect points which NBN Co intend to offer. It was also in this section that NBN Co further highlighted that for any given service point within a fibre serving area (FSA), only one PoI would be made available, either a local one (via the LEB product) or an aggregated one (via the AEB product). Based on my analysis so far, I must disagree fairly strongly with this intentional restriction of product offerings. Questions were also raised around thresholds for determining ‘contestable backhaul’.

My response argued that offering simultaneous and choose-able access to both an LEB and AEB product (wherever an AEB product is required due to lack of ‘contestable backhaul’) was highly desirable for maximum end-user choice and to support a healthy and competitive marketplace. In my examples I referred back to the local community services (and a consortium created or national specialised layer-3 wholesale RSP) as well as the power utilities (who already have existing backhaul network capabilities from pretty much every power transformer). Both of these cases represent clearly locally exclusive interests which would be commercially discriminated against by a forced AEB PoI. I also recommended that the threshold test for ‘contestable’ be set at three or more independent providers as opposed to just two (strongly hinted at). Besides, much of the negative impact and concerns around the determination of ‘unconstested’ backhaul’ vanishes if both local and aggregated access products are offered concurrently.

Section 7: NBN Co wholesale fibre bitstream products definition
In this section NBN Co provided much more detail around the product structure, in particular the different components involved in the LEB and AEB products. Each access product is really a bundle of an access link which is in effect the access service to the end-user premise, a connectivity link which is in effect the connection service provided at the PoI for each end-user access link and a transit link which represents the optional backhaul component of the service. The transit link is naturally omitted in the case of local PoI and present for any aggregated PoI. The remainder of the section then attempts to justify the decision not to offer the LEB and AEB products concurrently for a given end-user service access.

My response to this section introduced additional material to support concurrent delivery of LEB and AEB services in areas where ‘contested backhaul’ is deemed be unavailable. This analysis considered three exhaustive scenarios (including consideration of the null hypothesis that ‘contested backhaul’ exists of course, as in case 3) that demonstrated concrete market benefits from offering the two access products concurrently versus exclusively:

1. No market player has backhaul capacity to the local PoI. In this case NBN Co would construct backhaul capacity and offer an AEB product. All RSPs would voluntarily accept the AEB product at the aggregated PoI (having no other choice) with the exception of RSPs wishing to support exclusive local services (content caching providers, local community providers, power utilities, emergency services etc) who could accept the LEB product delivered at the local PoI. This represents a fair market whereas the exclusive approach would unfairly impact the RSPs supporting local services. Eventually it is hoped that such a market would lead to the next case.
2. One or two existing players have backhaul capacity to the local PoI. In this case NBN Co should construct/maintain backhaul capacity and offer an AEB product to any RSP that wishes to seek it in addition to the LEB product. Those RSPs who own the existing backhaul can choose to use their backhaul exclusively (via the LEB product), choose to abandon their backhaul (and seek the AEB product exclusively) or choose to partition their backhaul and enter the competitive backhaul market, offering LEB coupled with transit themselves to other RSPs. As in case 1 above, RSPs supporting local services can always seek the LEB product at the local PoI. This scenario results in optimal market behaviour whereas the exclusive approach forcefully deprecates the assets of existing backhaul providers and establishes a non-natural monopoly provider through NBN Co. Additionally the exclusive approach makes reaching the next case much more involved and difficult.
3. Sometime later, some previously ‘uncontested’ backhaul fibre serving areas will become considered ‘contested’. At this point in time all the existing backhaul providers can continue to decide how they wish to leverage their backhaul capacity, in addition NBN Co can negotiate a handover of transit capacity (and existing AEB services) to another market player or progressively withdraw the AEB product offering in this area, allowing for natural backhaul competition. Several, reasonably simple market led migration scenarios are possible for each of these cases. Under the concurrent approach, this scenario supports flexible and optimal market behaviours whilst the exclusive model forces a complex and potentially disruptive migration of services.

Maybe I’m being just a little melodramatic, yet I still believe that this decision warrants further engagement and consideration before it is dismissed.

Section 8: Important product elements
This was the last technical section of the document and arguably one of the most important. It was in this section that NBN Co introduced their current thinking around QoS (Quality of Service) differentiation models, support for traffic management and prioritisation, support for security and data integrity, support for voice services, multicast services and mention of possible support for an RF overlay. Questions within this section centered around the QoS model, the multiple service delivery intention and expectations around ports/interfaces (numbers and type), battery backup and other miscellaneous capabilities of the ONT (customer premise termination point).

My response to this section included significant detail on almost everything listed above with the exception of the IP multicast and RF overlay, both of which, I suspect others in the industry have much more to say about than do I. It is very good to see that NBN Co are seriously considering offering differentiated performance characteristics at differentiated prices. I’ve argued on this blog that such approaches are not only beneficial to the market place, but can even be wholly desirable for all players. Differentiated pricing is what makes industries such as airlines not only sustainable, but also affordable; it is the magic sauce that allows for social welfare discounts (think pensioner and student pricing) and without it, there would be no competitive automobile industry (and hence no vehicular choice). It is also entirely appropriate for natural monopoly telecommunications operators such as NBN Co, especially ones who are deploying optimal, shared infrastructures (as opposed to multiple overlays). I clarified some elements of their proposed architecture by defining four differentiated performance classes (Expedited, Assured, Premium and Standard) which can be provided via a mix of both access and PoI committed/peak rates. As part of this discussion I also highlighted some of the approaches that can be taken to pricing and dimensioning of these classes with particular attention to the ACCC (Australia’s competition watchdog).

Finally, my response to this section included further details about opportunities for synergies between NBN Co and the various utility providers (in particular electricity, but also gas, water and sewerage), recommendations for supporting voice services (in particular allowing end-user choice of preferred carrier for POTS) and some additional comments around battery backup approaches (namely user selectable and integrated into the ONT).

Section 9: Conclusion and next steps
There is not much to say about this section beyond merely closing on a positive note. It is very important that we (all Australian’s) make sure that we get the best network and hence the best services we can out of NBN Co. I have no doubts that the industry engagement model is a positive means to achieving this outcome. Together we can make sure that we get what everyone desires within reason. It would be a tremendous waste of $43 billion dollars if we were to get it wrong.

Advertisement