|
|
|
|
 |
|
|
|
|
|
|
|
|||||||||||||
 |
 |
 |
|
|||||||||||||||||
| Consultation on future interconnection arrangements for dial-up Internet in the United Kingdom November 2000 | |
|||||||||||||||||||
 |
 |
 |
|
|||||||||||||||||
 |
|
|
|
 |
|
|||||||||||||||
|
|
|
||||||||||||||||||
|
November 2000 This document is also available in pdf format. To view the document in pdf you will need to have Adobe Acrobat Reader If you already have Adobe Acrobat Reader you can download the document immediately. Chapter 4 Oftel’s preliminary conclusions and proposals Chapter 5 Further expansion of available tandem capacity Chapter 7 Double tandem conveyance Chapter 8 NTS interconnection links Annex 2 Single tandem FRIACO - charge calculation Annex 4 Access control services – a possible regulatory framework Annex 5 Glossary of terms and acronyms S.1 This consultation document proposes solutions – short, medium and long term – to ensure that there are effective interconnection arrangements to support expanding demand for Internet access over the Public Switched Telephony Network (‘PSTN’) in the UK. There is increasing demand for Internet access and the growing use of ‘flat rate’ (or ‘unmetered’) products will increase further the pressures on BT’s core network and switches. At present, around 99 per cent of residential consumers who access the Internet do so via PSTN dial-up. Broadband access technologies will grow, but it is likely that access to the Internet by dial-up over the PSTN will remain important for the foreseeable future. S.2 On 26 May 2000, the Director General of Telecommunications (the ‘Director General’) made a Direction requiring British Telecommunications (‘BT’) to provide a wholesale flat rate Internet access service, known as Flat Rate Internet Access Call Origination (‘FRIACO’). The service to be provided was transmission from the consumer to BT’s local exchanges (‘DLEs’) plus conveyance from the DLE to the Other Licensed Operators’ (‘OLOs’) Points of Connection (‘PoC’). In this respect, it is important to note that most OLOs have their PoCs at or near BT’s tandem switches. Tandem switches are main exchanges in the tandem network which act as single switching points for calls. S.3 The Direction of 26 May 2000 did not include a provision for a flat rate wholesale service between the DLE and the tandem switches. BT argued that the likely large increase in Internet traffic and call duration expected to result from flat rate products would place the tandem switches and the tandem network as a whole under severe strain and could cause significant operational difficulties for both voice and data traffic. Given the Director General’s duty to ensure that all reasonable demands from consumers are met, he could not take action which might jeopardise this. Therefore, he commissioned a panel of experts (the ‘Panel’) to examine the issue. The Panel’s report has been considered by Oftel and taken into account in Oftel’s conclusions and proposals set out in this consultation document. S.4 The Panel found that metered (or ‘pence-per-minute’) Internet traffic is doubling about every 10 months and the total of all traffic over the tandem network will double within two years. The Panel further noted that the introduction of unmetered Internet tariffs would significantly increase total traffic, since unmetered call durations are, on average, four times longer than metered call durations. The Panel concluded that changes need to be made to the present network architecture to ensure sustainable and efficient long-term arrangements. It also made proposals for short and medium term action, including removing some Internet traffic from the tandem network and adding further capacity, to ensure that traffic growth does not overwhelm the capability of the tandem network. S.5 Oftel’s aims are fair competition in Internet access and network arrangements that secure the robust provision of both data and voice services. Oftel considers that BT is dominant in conveyance up to and including the tandem switch. Therefore, regulation is appropriate to ensure that BT provides such conveyance to OLOs at a price which approximates to that which would be set in a competitive market. Oftel’s Proposals S.6 Oftel proposes that there should be the following obligations on BT:
S.7 BT’s obligation under the FRIACO Direction of 26 May 2000 to provide unmetered wholesale Internet call origination to the DLE will continue. Short term (Stage 1) – (1 February 2001 – 31 January 2002) S.8 In the short term, Oftel accepts that if unmetered traffic grows as estimated, there could be a shortage of tandem switch capacity. Therefore, Oftel considers that a way needs to be found to ration such switched capacity so that BT’s obligation to provide single tandem FRIACO up to the ceiling can be met. S.9 Oftel considers there are two main options for rationing of tandem switch capacity: Option 1 Option 1 requires OLOs purchasing single tandem FRIACO to initiate the necessary changes themselves. It would place an obligation on OLOs who wish to purchase single tandem FRIACO to take their Internet traffic off the PSTN at those DLEs which are co-located with tandem switches where the OLOs have a PoC. At present, some 340 DLEs are co-located at tandem switches. Oftel would propose that this process should start with the lowest cost changes – that is, changes in respect of those OLOs who have built out their own networks to these tandem switches with co-located DLEs. Only if necessary would this obligation on OLOs extend to DLEs co-located at tandem switches to which OLOs connect via Interconnection Extension Circuits (‘IECs’) and to non-co-located DLEs. The extent of the requirement (that is, the volume of traffic to be taken off) would depend on the amount of additional capacity at the tandem switch (or ‘headroom’) it is necessary to create. Oftel proposes that the necessary changes, which involve network re-configuration, should be made by the end of April 2001. OLOs which are required to make such changes would continue to pay the standard single tandem interconnection charge fixed under the Network Charge Control (‘NCC’). Oftel would propose that the costs of traffic re-arrangement, including for example additional links, should not be recovered by BT using additional charges. However, such costs would be included in the forecast cost base for standard interconnection charges under the NCC by adding them to the costs of the tandem switch component. BT would remain under an obligation to provide tandem switch interconnection for all traffic from all other DLEs. Option 2 Option 2 would require BT to initiate the necessary changes. As stated above, BT is to remain under an obligation to convey all traffic (that is, data and voice) from DLEs to the OLOs’ PoCs at the tandem switch premises. It would be up to BT to determine how such conveyance could be achieved, either through the switch as currently, or by alternative means such as taking the traffic off the PSTN at co-located DLEs and subsequently conveying traffic to the OLOs’ PoC, but not through the tandem switch. In the latter case, OLOs purchasing single tandem FRIACO would be required to accept re-arrangement of their traffic. Regardless of the means of conveyance, OLOs would pay the standard charge for single tandem interconnection, fixed under the NCC. As under Option 1, costs of any traffic re-arrangement would be included in the forecast cost base for standard interconnection charges. S.10 Neither Option 1 nor Option 2 prevents an OLO deciding voluntarily to build out its own network to the DLE. In this case the OLO would pay standard DLE interconnection charges and also bear the cost of any traffic re-arrangements. S.11 Oftel favours Option 2, because it places strong incentives on BT to minimise costs to create headroom as its revenues will remain constant irrespective of the method of conveyance. It also enables traffic re-arrangement to be adjusted according to the actual growth of traffic on the tandem network, rather than the more rigid set of requirements under Option 1, which would depend on Oftel’s forecasts. S.12 Oftel proposes to implement the obligations under either Option 1 or 2 by 1 February 2001. A choice will have to be made, as Options 1 and 2 cannot co-exist. To avoid overload on the tandem switches, BT’s obligation to provide single tandem FRIACO will be capped by a ceiling initially proposed at 13,500 tandem switch 2Mbit/s ports. From 1 May 2001, Oftel expects to raise the ceiling significantly. Thereafter, the ceiling will be reviewed periodically and will be removed entirely by 31 January 2002. Medium term (Stage 2) – (1 February 2002 onwards) S.13 From 1 February 2002, Oftel proposes that the ceiling on BT’s obligation to provide single tandem FRIACO should be removed. The requirements on OLOs in respect of traffic re-arrangement would also be removed. Therefore, in parallel with the implementation of Stage 1, Oftel expects BT to take steps to be able to meet this obligation from 1 February 2002. BT may need to make new investments to expand capacity. One possibility is an ISP tandem switch overlay network as identified by the Panel. This would involve BT adding further tandem switches to the tandem network. However, Oftel accepts that there is a risk that BT may be unable to recover the cost of such investment. This is because OLOs might only use such new switched capacity for a short period before moving to more efficient alternatives such as IP-based interconnection. In Oftel’s view, the best approach to this issue is for the risk of any new investment to be shared between BT and the OLOs. Proposals for such risk-sharing are discussed in this document. Longer term (Stages 2 and 3) S.14 In the longer term, conveyance of Internet traffic through the tandem switch layer on a circuit-switched network (that is, the PSTN) is unlikely to be the most efficient option for narrowband Internet services. OLOs may choose to build out to more DLEs, but building out to all DLEs is unlikely to be cost effective for each and every OLO. The Panel strongly recommended that IP interconnection would be the best long-term solution, because it considered IP networks are more efficient than circuit-switched networks for Internet traffic. Such interconnection would enable BT to ‘groom’ (in other words, take off the PSTN) Internet traffic at DLEs, using BT’s modems and onto BT’s IP network with which OLOs would interconnect. S.15 Oftel shares the Panel’s views in respect of the probable advantages of IP interconnection in terms of efficiency and lower cost. However, Oftel notes that there are both technical and regulatory issues that need to be addressed. In particular, in respect of IP interconnection, the log-in process and the consumer data it generates would be under BT’s control. This raises concerns about possible anti-competitive behaviour. Further work is needed to determine the appropriate regulatory response. The Access Control Services regime for interactive services may be a helpful model. The nature of any regulation will need to be based on a market analysis. Therefore, Oftel proposes to set up industry groups, which it would chair, to take forward the regulatory and technical issues associated with IP interconnection. In Oftel’s view, work should commence now to widen the range of interconnection products available to include IP interconnection for Internet access originating on the PSTN. Consultation S.16 Oftel seeks comments on these proposals by 8 December 2000. Thereafter, Oftel will receive further comments on comments until close on 22 December 2000. S.17 The Director General intends to formalise the issues outlined in this document in the form of a draft Direction to be issued shortly. Comments will be sought accordingly. S.18 The table below sets out Oftel’s proposals in outline: Oftel’s proposals
Introduction 1.1 This document proposes solutions to alleviate pressures on BT’s tandem network caused by increased use of the Internet. The recent introduction of ‘flat rate’ (or ‘unmetered’) Internet products, which typically increases the time users spend on-line, is set to amplify these pressures. The solutions proposed take full account of the report (the ‘Report’) prepared by an independent panel of experts (the ‘Panel’), chaired by Professor John O’Reilly of University College London, a copy of which can be found at Annex 1 of this document. The document is primarily concerned with dial-up Internet access; that is access to the Internet over the Public Switched Telephony Network (‘PSTN’) using a computer and a modem. Internet access via broadband, such as BT’s ADSL service or other DSL services provided by OLOs through ‘local loop unbundling’, does not use the tandem PSTN and, hence, does not contribute to these pressures. Indeed, broadband access may help alleviate the pressures if and when it is substituted for dial-up access. Oftel’s objectives 1.2 In approaching these issues, Oftel has been guided by its overall goal of the best deal for the consumer in terms of choice, quality and value for money. Oftel wishes to see effective competition in dial-up Internet access which, Oftel recognises, is likely to continue to be the means by which most residential consumers access the Internet in the foreseeable future. Oftel also seeks effective competition in basic telephony services. Therefore, Oftel proposes robust arrangements which will promote and facilitate future competitive developments, but which also safeguard existing services in a cost-effective manner. 1.3 The issues involved are complex. Achieving the objectives will require efforts by all parties, including BT, competing Other Licensed Operators (‘OLOs’) and Oftel. Oftel’s proposals set the regulatory framework for new arrangements. Nevertheless, there is much work to be done by the industry, not least to re-configure existing networks, and possibly to build new network capacity. This requires effort, commitment and co-operation between interconnecting operators. Oftel will not hesitate to take action where appropriate to bring this about. Background 1.4 On 24 December 1999, Oftel received from MCI/WorldCom (‘MCIW’) a complaint. This concerned BT’s refusal to offer a flat rate interconnection product for Internet access call origination with call handover at DMSUs (in other words, MCIW requested ‘single tandem FRIACO’). A DMSU (or ‘Digital Main Switching Unit’) is the principal type of tandem switch and, unless stated otherwise, this document concerns tandem switches generally (tandem switches are main exchanges in the tandem network which act as single switching points for calls). 1.5 In its response to the complaint, BT submitted that the expected substantial increase in traffic, which would result from unmetered interconnection for Internet purposes, would exceed the planned capacity of the PSTN, as currently configured, by mid 2001. 1.6 BT’s view was that, despite an extensive tandem switch replacement programme, which BT was already undertaking, there was a significant risk that requiring connection for unmetered Internet traffic, which would be switched through the DMSU, would cause significant operational difficulties. Since DMSUs, and their replacement New Generation Switches (‘NGS’ – a type of new high capacity tandem switch) carry both voice and data traffic, BT argued that the potential consequences of capacity difficulties would be far reaching. To avoid this, a significant additional investment in switch capacity would have to be made. 1.7 Oftel sought independent expert advice from the Panel in respect of BT’s arguments. Until this was received, the Director General could not determine whether MCIW’s request for a flat rate, unmetered interconnection product for Internet access call origination with call handover at the DMSU was reasonable. 1.8 Nevertheless, on 26 May 2000, the Director General determined that Flat Rate Internet Access Call Origination (‘FRIACO’) at the local exchange (‘DLE’) was reasonably required and, hence, he directed BT to provide it. The Director General also directed BT to provide conveyance from DLEs to Points of Connection (‘PoC’) of OLOs at tandem switch premises (any reference in this document to interconnection "to" or "at tandem switch premises" or "DMSU premises" will be taken also to apply to OLOs’ PoCs where they are located outside BT’s tandem switch premises). A copy of the Direction of 26 May 2000 can be found at: http://www.oftel.gov.uk/competition/fria0500. As a result, BT made FRIACO at the DLE available from 1 June 2000. 1.9 Shortly afterwards, BT and MCIW resumed discussions on how FRIACO could be conveyed to the DMSU premises. Following these discussions, on 3 August 2000, BT announced the availability of FRIACO ‘Hybrid’ from 1 September 2000. This enabled OLOs to purchase FRIACO at the DLE with onward conveyance to the DMSU at normal ‘local-to-tandem’ metered conveyance rates. This product complements the original FRIACO, as determined on 26 May 2000, since it enables OLOs to gain access to FRIACO traffic from their DMSU PoCs by paying pence-per-minute charges for DLE to DMSU conveyance, in addition to the unmetered charge for FRIACO to the DLE. 1.10 Consistent with BT’s view that the DMSUs will not be able to handle the expected increase in traffic arising from the introduction of unmetered Internet traffic, BT has taken steps to ‘groom’ (in other words, take off the PSTN) its own Internet traffic at ‘enabled’ DLEs; that is, DLEs that have ‘grooming capability’ (capability to take traffic off the PSTN onto the IP network). Such traffic is terminated on its IP network. BT first outlined plans to roll-out its IP network nearer to the customer and so groom Internet traffic at the DLE, at the Oftel Internet Forum (‘OIF’) on 16 November 1999. BT’s press release of 17 November 1999 can be found at: http://www.bt.net/news_press_expansion.htm. 1.11 Immediately after BT announced its intention to launch SurfTime on 7 December 1999, Oftel called another OIF meeting which took place on 21 December 1999. At this meeting, BT presented its more detailed plans regarding SurfTime, which was to be launched in Spring 2000. Subsequently, BT developed SurfTime as an unmetered Internet call origination product which conveys traffic from the customer to the DLE. BT has made clear that the network architecture underlying its own retail SurfTime operation via BT Internet involves grooming at the DLE. 1.12 Following the announcement of SurfTime, there were extensive discussions, facilitated by Oftel, between BT and OLOs about the appropriate interconnection arrangements, in recognition of the fact that the existing PoCs for most OLOs are at the DMSU premises. The outcome was a DMSU-DLE migration plan, whereby OLOs could interconnect with BT’s SurfTime, using their own DMSU PoCs, on condition that they made arrangements to reduce the volumes of unmetered Internet traffic at the DMSU to 33 per cent by the end of September 2000 and to 25 per cent by the end of December 2000. Continued migration after 31 December 2000 was to be reviewed by BT together with OLOs in September 2000 (to date, this review has not been conducted). For the full details of these arrangements, see paragraph 5.4.8 of Annex A to BT’s Network Charge Control Standard Interconnect Agreement (October 1997), edition July 2000, available on www.btinterconnect.com. 1.13 BT’s product description for FRIACO ‘Hybrid’ states that "it is an interim solution until 31/12/2000. The arrangements to apply from 1/1/2001 will be reviewed in September 2000 (the same timeline as applies to the review of OLO arrangements made in the SurfTime case)". FRIACO ‘Hybrid’ is not restricted to the same DMSU to DLE migration percentages as SurfTime, but will be subject to whatever arrangements will be agreed with BT.
The report 2.1 The Panel’s report has been reproduced at Annex 1 of this document (where necessary, confidential information has been removed.) 2.2 Oftel asked the Panel to consider options for the interconnection of unmetered Internet traffic at the tandem switch premises. As noted above, BT had expressed concerns about the practicability of expanding the present network architecture of switching via DMSUs. Pending the Report, the FRIACO Direction of 26 May 2000 did not address the issue of interconnection via the DMSU or other tandem switches. 2.3 However, the Direction did state that interconnection at the DMSU premises was reasonable. Therefore, it required BT to provide conveyance to OLOs’ PoCs at DMSU premises. Subsequently, BT offered this product using IECs. 2.4 OLOs did not consider it reasonable or practical to connect to every DLE, even if the capacity were extended to the DMSU premises. They commented that interconnection at tandem switches could avoid inefficiencies because of the concentration function at such tandem switches. In other words, there was still a considerable demand for a tandem switched unmetered product. Being a partly metered product, FRIACO ‘Hybrid’ does not meet that demand. 2.5 The Panel noted that Internet traffic is currently doubling every 10 months and that this growth is expected to increase even more steeply once unmetered services are introduced. Such unmetered services could increase holding times by possibly a factor of four and, in general, they are also likely further to stimulate the underlying Internet growth. Taking account of BT’s present expansion plans for its tandem network, the Panel believes that urgent action is needed to ensure that such traffic growth does not overwhelm the capability of the tandem network. In broad terms, this can be achieved by finding ways to remove the Internet traffic from the tandem layer or by adding further tandem capacity in a way that is practical and sustainable. 2.6 The Panel takes the view that the present architecture of ‘meshed’ DMSUs is the least suitable for Internet traffic once unmetered services become widespread (each of the present 74 DMSUs is ‘fully meshed’ (or interconnected) with all other 73 DMSUs.) This is because, beyond a certain number of switched nodes (currently 74), the costs of such a fully-meshed network rise sharply and, even with the present major investment in high capacity NGS, the network would soon become both expensive and unmanageable. Such a network could be overwhelmed by a rapid increase in demand for Internet access. 2.7 The Panel takes the view that, ideally, Internet traffic should be taken off the PSTN at the earliest point in the network. However, given that not all OLOs can reasonably connect to all of around 740 ‘enabled’ DLEs (as stated above, any reference in this document to interconnection "to" or "at tandem switch premises" or "DMSU premises" will be taken also to apply to OLOs’ PoCs where they are located outside BT’s tandem switch premises), some form of traffic concentration is necessary. Although the Panel has analysed a number of variants for delivery of such capability, its strongly expressed view is that, in the longer term, an interconnection product based on IP technology would be the most efficient, taking into account costs, manageability and synergy with future broadband, non-dial, call origination methods, such as ADSL and cable modems. 2.8 Such IP products are to a degree already available as retail products from BT, but they do not meet all the needs of OLOs to specify their own service parameters and maintain control of the customer relationship. The Panel concluded that this option could not be specified and implemented in time to resolve capacity issues. 2.9 In recognition of this, the Panel considered the following interconnection product options. Options B to E are PSTN interconnection products:
2.10 The Report addresses the pros and cons of these options. Option B, relying on the continued expansion of the DMSU mesh, is considered to be the least desirable solution for the reasons set out above. Option C, which is already part of the existing FRIACO product, suffers from the difficulty that, as discussed above, few OLOs wish to reach all DLEs and the fact that there are significant inefficiencies from a large number of small routes. Option D is seen as workable, but requires new tandem capacity to be added to the BT network, which will take time to install and configure. 2.11 The Panel believes that Option E may provide some short-term additional capacity (or ‘headroom’) but that it is not suitable for longer-term use, as it would rely on expansion of the tandem switch layer. However, the Panel feels that Option E would provide sufficient capacity headroom to allow for Option D to be introduced. 2.12 Therefore, in conclusion, the Panel recommends that Options E and D should be implemented. Option E should provide sufficient capacity headroom to allow time for Option D to be implemented. Furthermore, the Panel recommends that work should start immediately on Option A, the IP interconnection product, as the Panel expects that there will be strong economic and technical pressures for this, in particular, to provide synergy with broadband IP.
Oftel’s approach Timescale 3.1 The Panel’s Report makes it clear that, in the near future, the current network architecture will be unable to cope with the estimated growth of unmetered Internet traffic. According to BT’s own forecast, if no action is taken, the total volume of all traffic, including voice and data, will put constraints on network capacity and could reach a critical stage by mid-2001, unless traffic is off-loaded or new capacity is provided (see Annex F of the Report, page 59) Whilst noting the Panel’s view that broadband Internet access will soon become significant, Oftel also notes that, at present, around 99 per cent of residential consumers who access the Internet do so via dial-up over the PSTN. While in future broadband access technologies will grow, it is likely that access to the Internet by dial-up over the PSTN will remain important for the foreseeable future. Even in the US, it is estimated that around 90 per cent of residential consumers who access the Internet do so via dial-up access. Therefore, a solution that allows mass market take-up of unmetered services via dial-up over the PSTN is very important. 3.2 Oftel considers it essential that it acts now in order to ensure the uninterrupted provision of telephony services and to allow the widespread introduction of unmetered Internet traffic by both BT and OLOs. Given the time constraints imposed by BT’s estimated growth in Internet traffic, Oftel considers it appropriate to instigate a three-stage solution (see Chapter 4 below). 3.3 However, before doing so, Oftel sets out its views on how it proposes to approach the issue of the appropriate regulatory framework for interconnection of Internet traffic. Dominance in Single Tandem Call Origination 3.4 An assessment of market power is an important factor in determining the appropriate level of regulation. The Direction of 26 May 2000 reflected Oftel’s view that BT is dominant in call origination up to DLE level. Oftel considers that BT is also dominant in conveyance between DLEs and tandem switches. These two services together comprise single tandem call origination; that is, from the customer, through the DLE and tandem switch to an OLO’s PoC. This finding is reflected in the operation of the Network Charge Controls (‘NCC’) where call origination and DLE conveyance are specified to be non-competitive services and are included in a basket which is subject to a RPI-8% control. In its price control review proposals, published on 27 October 2000, Oftel has set out its view that these services should remain in a RPI-x basket in the forthcoming new NCC. RPI-x is the system of price control where average annual price changes for the price-controlled services are limited to the increase in inflation (as measured by the charges in the Retail Price Index (‘RPI’)), less a specified factor of x. 3.5 Pursuant to Article 4(1) of European Parliament and Council Directive 97/33/EC (the ‘Interconnection Directive’ or ‘ICD’), BT is required to offer to enter into an agreement with an ‘Annex II’ operator to connect that operator’s telecommunications system to the BT telecommunications system through PoCs of sufficient number and capacity so as to conveniently meet all reasonable demand. BT is also required to provide such other telecommunications services as are reasonably required to enable BT to fulfil this obligation. 3.6 On the basis of both BT’s dominance and its obligations pursuant to the ICD, Oftel believes that BT should have an obligation to provide call origination to the tandem switch premises. In addition, Oftel believes that BT should have an obligation to provide unmetered Internet access, single tandem call origination through the tandem switch (‘single tandem FRIACO’), but this obligation should be limited, to the extent that it is technically feasible, for a period of time so that further capacity can be provided. 3.7 At present, most OLOs interconnect at the DMSU switch. OLOs may build out to DLEs if they wish, but Oftel’s view is that it is unrealistic to expect every OLO to interconnect with all of around 740 of BT’s ‘enabled’ DLEs. In other words, Oftel believes that, where it is technically feasible, traffic concentration at a tandem switch is a reasonable request and that OLOs should not have to replicate BT’s network to every DLE. 3.8 BT’s present obligation to provide interconnection for metered Internet traffic, pursuant to its Reference Interconnection Offer, remains unaffected.
Oftel’s preliminary conclusions and proposals 4.1 At this stage, on the basis of the information presently available and for the reasons set out in Chapter 6, Oftel believes that the Panel’s Option B is not a technically feasible option and that its Option C, on its own, is not sufficient to meet reasonable demand. Therefore, the approaches discussed in this chapter are based on the Panel’s Options A, D and E. 4.2 Oftel’s preliminary conclusion is that BT should be required to provide single tandem FRIACO. Initially, to ensure that this obligation is reasonable, it would be subject to a ceiling on the volume of FRIACO services required to be supplied and to some specific requirements on OLOs purchasing single tandem FRIACO. Subsequently, the ceiling would be removed and the requirements on OLOs would cease to apply. This chapter sets out Oftel’s proposals to ensure that reasonable demand from OLOs can be met and to achieve the objectives discussed in Chapter 1. 4.3 Oftel proposes that the charge for single tandem FRIACO should be £17,692.32 per 2Mbit/s tandem switch port (or £589.74 per 64Kbit/s). This charge has been derived by applying the same methodology as that used in the FRIACO Direction of 26 May 2000. Details of the calculation of the single tandem FRIACO charge are set out in Annex 2 of this document. 4.4 As discussed above, Internet traffic is expected to continue to grow very rapidly and the anticipated increase in unmetered Internet is likely to be accompanied by a significant increase in the duration of calls. Oftel believes that the demand for tandem switch capacity arising from such traffic will not be supported by a continuation of BT’s present network architecture and existing plans to expand capacity. This view is based on the Panel’s conclusions as set out in the Report. 4.5 Specifically, Oftel considers that what is needed in the short term (that is, up to 31 January 2002) is a method of rationing tandem switch capacity, because BT’s existing network is unable to cope with provision of interconnection for all metered and unmetered traffic at the tandem layer. In parallel with such rationing, it is likely that steps need to be taken by BT to ensure that more capacity at tandem switch premises is made available in the medium term. Finally, some steps need to be taken to ensure a set of viable longer-term solutions. 4.6 Having taken account of the Panel’s conclusions and recommendations in the light of the relevant economic, legal and regulatory factors, Oftel proposes a three-stage process to ensure that its objectives, as set out in Chapter 1 above, are met. Stages 1 and 2 are based on, respectively, the Panel’s Options E and D, whilst Stage 3 is based on the Panel’s Option A. 4.7 It is important to bear in mind that the conclusions and proposals set out in this consultation document are preliminary and subject to receiving additional estimates, figures and other information from BT as well as comments from the rest of the industry. Oftel has already requested BT to respond to a number of questions regarding, for example, capacity, traffic increase and volume forecasts and PoCs. Oftel would welcome OLOs’ views on the type of questions that they believe need answering in this respect. Once all the responses have been received, Oftel will scrutinise the information carefully. Oftel welcomes any additional relevant information OLOs are able to provide. Stage 1: Interconnection at DLEs (from 1 February 2001) 4.8 On the basis of the Panel’s Report, Oftel believes that, if unmetered Internet services are widely adopted, there is likely to be a shortage of tandem switch capacity and that rationing of such capacity is required in the short term. Such rationing can be achieved by migrating traffic away from tandem switch interconnection to DLE interconnection, which would create additional capacity (or ‘headroom’) at the tandem switch. In this respect, the main questions that arise are: Which traffic should be re-arranged from tandem interconnection to DLE interconnection? Who should pay for the costs of the traffic re-arrangements? Both questions are discussed below in the context of two different options for the creation of the necessary headroom. Stage 1, Option 1: Mandated Interconnection at DLEs Which Traffic? 4.9 Option 1 would place a requirement on OLOs, who are purchasing single tandem FRIACO, to re-arrange traffic from tandem switch to DLE interconnection in specified circumstances. 4.10 The specification of the relevant circumstances would start from the premise that the resource, that is tandem switch capacity, should be available to those OLOs who value it most. In other words, those who value it least should be supplied last. The value of tandem switching to an OLO depends on the cost of the next best alternative. This cost/value would tend to be lower for an OLO which is able to migrate traffic from an existing PoC at a tandem switch to DLEs that are co-located at the same site. At present, there are around 340 ‘enabled’ DLEs that are co-located at 61 of BT’s tandem switch premises. However, the cost of traffic re-arrangement is likely to differ materially, depending on the way in which the particular OLO’s PoCs have been established. 4.11 An OLO might have built out its network and established its PoCs using In-Span Interconnect (‘ISI’). In this case, some re-arrangement costs might arise if there was a need to establish signalling links to the Signalling Transfer Point (‘STP’). The thinner interconnect routes at co-located DLEs might result in the OLO facing additional costs on its own network, such as the use of additional switch ports and the costs of managing a larger number of interconnect routes. There would also be the cost arising from the provision by BT of the necessary Intra-Building Links (‘IBL’) to establish the PoCs at the co-located DLEs. The Erlang inefficiencies associated with thinner routes (that is, lower use) would be suffered on the IBLs. But, this effect is unlikely to be large because of the short length of the IBLs. 4.12 However, the additional cost arising from Erlang inefficiency might be much more significant if the OLO’s PoC is established using IECs or Customer-Sited Interconnect (‘CSI’). Such interconnection circuits are used by OLOs where they have not built out their own networks to the tandem switch premises. In these circumstances, the Erlang inefficiency of thinner interconnect routes would be suffered over the length of the IECs/CSIs, which might not be short links and, in any event, would be longer than IBLs. 4.13 Therefore, a ranking can be established for the value placed on tandem switch capacity by OLOs who are purchasing single tandem FRIACO. The lowest value would be for traffic from DLEs which are co-located at premises where the OLO has an existing PoC with a tandem switch, using ISI. The next lowest value would be for traffic from co-located DLEs, at a site where the OLO has IEC/CSI interconnection. The value would be higher for traffic from DLEs which are not co-located at tandem switch premises but are connected to tandem switches at premises where the OLO has a PoC, since the OLO would be required to obtain IECs or build out its own network to the DLEs. 4.14 The requirement on OLOs to re-arrange traffic to DLE interconnection would apply as a condition of purchasing single tandem FRIACO. The extent of the requirement on such OLOs would depend on the amount of headroom required. The types of traffic that might be required to be re-arranged would include wholesale unmetered Internet traffic; that is, FRIACO. But it might also include metered data traffic and voice traffic, if required to create sufficient headroom. The mandated traffic re-arrangement would include the lowest value category, that is ISI at co-located DLEs, as discussed above. If necessary to create the required headroom, it would also include higher value categories, IEC interconnection at co-located DLEs, and interconnection at DLEs which are not co-located. The exact nature of the requirements under this approach would depend on the result of the further detailed analysis discussed in paragraph 4.7. 4.15 An OLO purchasing single tandem FRIACO on a date soon after the final Direction, should be allowed a period of time to satisfy the specified requirements to interconnect at DLEs. Oftel would envisage that this period should be three months from the date of the final Direction which, subject to comments received during the consultation period, Oftel aims to publish in January 2001. If an OLO wished to place an order for single tandem FRIACO after the end of that period, it should satisfy the specified requirements for DLE interconnection as soon as it starts to purchase single tandem FRIACO. Moreover, if an OLO purchasing single tandem FRIACO were to choose to establish additional tandem switch PoCs in the future, it would be required to interconnect with DLEs co-located at those tandem switch premises as well. Oftel would propose that these requirements on OLOs should continue until 31 January 2002, after which date they would cease to apply. Who pays? 4.16 The OLO could be required to pay BT for the IBLs, STP links, IECs or other links used as part of the traffic re-arrangement. However, this approach might be considered inefficient and inequitable. The need to create headroom arises because, at least in Stage 1, the demand for tandem switch capacity from all sources may exceed the available supply. Therefore, the creation of headroom tandem switch capacity would benefit all users of such capacity, whereas the OLOs purchasing single tandem FRIACO would be bearing disproportionate costs of creating that headroom. Therefore, it would be preferable for all users of tandem switch capacity to bear a fair share of these costs. To achieve this, the costs to BT of providing such services could be forecast and included in the cost base for standard interconnection charges, when the forthcoming NCC is set in January 2001. In particular, the forecast costs could be added as an ‘uplift’ to the costs of the tandem switch component (‘main and digital junction switching’). With this approach, BT would be levying no specific charges on OLOs for the provision of IBLs, STP links, IECs or other links in relation to the mandated traffic re-arrangement. 4.17 The need to create tandem switch headroom means that, under Option 1, OLOs purchasing single tandem FRIACO would be mandated to accept traffic re-arrangement, that is, a substitute for tandem switch interconnection, comprising DLE interconnection plus IBLs, STP links, IECs or other links (for which it would pay no specific charges). Nevertheless, since the OLO has chosen tandem switch interconnection, Oftel considers it appropriate that the OLO should continue to pay single tandem charges. In other words, the OLO’s PoC would physically be interconnected to a DLE, but the OLO would pay single tandem (and interconnection circuit) charges, as if interconnection were to a tandem switch. 4.18 The OLO’s ‘build-or-buy’ decision, which is an important feature of the interconnection regime, would be retained. The OLO could choose to interconnect at the DLE, in which case it would have to pay for the IBLs, STP links, IECs or other links, but it would pay reduced interconnection usage charges at the local exchange segment rate (either call origination or call termination charges, depending on the nature of the traffic), rather than single tandem. 4.19 Another reason why the OLO should pay single tandem where traffic re-arrangement is mandated is that this would avoid the effect that the OLOs purchasing single tandem FRIACO would not be paying a fair share of the costs of the IBLs, STP links, IECs or other links used in the re-arrangement. As discussed above, the most natural way for these costs to be recovered from all users of tandem switch capacity would be to apply an ‘uplift’ to the tandem switch component. However, if the OLOs whose traffic had been re-arranged were only paying charges at the local exchange segment rate, they would be making no contribution to the recovery of such costs (on the re-arranged traffic). This problem would be avoided if the OLOs were to pay at the single tandem rate, as suggested above. 4.20 When setting the NCC, Oftel would take into account all of the payment arrangements described above. BT would receive single tandem payments for the mandated traffic re-arrangements, even though it would be physically providing and incurring the costs of the local exchange segment. This would partially offset the increase in the cost base caused by the inclusion of the costs of IBLs, STP links, IECs or other links used in the mandatory traffic re-arrangement. It is important to note that both effects would be taken into account on a forecast basis; that is, once the new NCC is set, there would be no pass-through of higher costs into higher charges, or higher revenues into lower charges. 4.21 It appears unavoidable that OLOs would bear any additional costs that would arise on their own networks from the implications of thinner routes. Stage 1, Option 2: Re-arrangement initiated by BT 4.22 Under the second option for creating the short term headroom, BT would initiate the re-arrangement of traffic to DLE interconnection. Rather than an OLO who purchased single tandem FRIACO being mandated to re-arrange traffic, as in Option 1, it would face a different requirement. In Option 2, such an OLO would be required to accept the re-arrangement of FRIACO, other data or voice traffic initiated by BT. In other words, the OLO’s consent would not be required. However, the OLO would face no specific charges for the provision of IBLs, STP links, IECs or other links which would be used to achieve the traffic re-arrangement and to convey the traffic to the OLO’s PoC. 4.23 Some of the features of Option 2 would be very similar to those of Option 1. An OLO would make the same interconnection payments to BT irrespective of whether the re-arrangement had been initiated by BT. In other words, the OLO would pay single tandem (and interconnection circuit) charges, as if interconnection were to a tandem switch. OLOs would also retain a ‘build-or-buy’ decision. Further, the implications for the NCC would be similar. Oftel’s Preliminary Conclusions 4.24 Although many features of Options 1 and 2 are similar, Oftel favours Option 2 because it has a number of more desirable characteristics. First, Option 2 would allow the decision about the required amount of headroom to be made flexibly by BT in response to the developing situation. Under Option 1, the decisions would be made in advance on a forecast basis by Oftel, which would unavoidably be subject to a potentially large margin of error, given the difficulties of developing a robust forecast. The greater flexibility of Option 2 would improve the likelihood not only that sufficient headroom will be created in aggregate, but also that it would be created at the particular tandem switches where the demand exists. 4.25 Second, BT would face good incentives to fulfil its specified obligation to provide tandem switch interconnection capacity in the way that would minimise resource costs. This is because, when making its decisions, BT would be comparing the marginal costs of the different possibilities for creating headroom, which would also include re-arranging its own traffic, and so BT would have an incentive to choose the least cost method. 4.26 In the case of FRIACO traffic, there is an additional consideration, which BT would have to take into account when deciding how to re-arrange traffic. If single tandem FRIACO were provided by BT using DLE interconnection plus IBLs, STP links, IECs or other links to the tandem premises (that is, as re-arranged traffic) it might need to supply more capacity than the OLO had paid for. This arises from the indivisibility in FRIACO port capacity (following from the fact that capacity is purchased in 2Mbit/s ‘chunks’). 4.27 An OLO cannot exceed the number of minutes that can be carried through the FRIACO port, because FRIACO ports are physically separate from other ports. If an OLO were to purchase single tandem FRIACO, physically provided through the tandem switch, this limitation on the total minutes would be determined by the FRIACO port at the tandem switch. However, if the request for single tandem FRIACO were satisfied using physical interconnection at the DLE plus IBLs, STP links, IECs or other links, there would be no physical FRIACO port at the tandem switch. Instead, the limitation on traffic would be provided by FRIACO ports at the DLE. When re-arranging traffic, BT should provide at least as much traffic-handling capacity as would be available to the OLO if the physical interconnection were at the tandem switch. Given the indivisibility in the supply of FRIACO ports in 2Mbit/s ‘chunks’, this might result in the OLO receiving more FRIACO port capacity in total. 4.28 This point can be illustrated by using a simple numerical example. Assume that the OLO would have 2Mbit/s of traffic filling a single FRIACO port at the tandem switch (for ease of exposition, the unrealistic assumption is made that Erlangs per circuit would be unity). Assume that this comprised 0.5Mbit/s of traffic originating on one DLE and 1.5Mbit/s on another DLE. If the single tandem FRIACO service were provided by BT using DLE interconnection plus IBLs, STP links, IECs or other links (in other words, re-arranged traffic), BT would need to provide a FRIACO port of 2Mbit/s at each DLE. Hence, it would be providing the OLO with FRIACO capacity of 4Mbit/s. But, the payment by the OLO would continue to reflect the number of FRIACO ports it ordered at the tandem switch, which would be 2Mbit/s in this simple example. At higher traffic volumes the implications of the indivisibility would be smaller. This feature of re-arranging FRIACO traffic would be part of the cost faced by BT when deciding which traffic would be the least cost to re-arrange to DLE interconnection. 4.29 The advantages of Option 2 arise from the flexibility that it gives BT to initiate traffic re-arrangement. Although the incentives faced by BT when making such decisions are generally sound, there might be some potential to use the flexibility in an anti-competitive manner. Oftel proposes to consider any allegations of anti-competitive behaviour by BT on a case-by-case basis. 4.30 Oftel proposes that the requirements on OLOs under Option 2 should continue until 31 January 2002 and then cease to apply. The above discussion of limitations on FRIACO minutes points to an advantage arising from the concentration of traffic at tandem switches, in particular in the context of FRIACO, because of the indivisibility in port capacity. Other advantages of tandem switch interconnection include the saving of costs of IECs or other links and costs incurred on OLOs’ networks. These factors suggest the desirability of removing the requirements for DLE interconnection at the beginning of Stage 2. Ceilings on tandem switch capacity 4.31 A possible conclusion of Oftel’s further analysis, outlined in paragraph 4.7, is that, under either approach, the extent of the requirements to limit demand for tandem switch capacity might not be sufficient to create the headroom tandem switch capacity necessary to satisfy all demands for tandem switch capacity. For this reason, Oftel proposes to establish a ceiling on the amount of tandem switch capacity that BT would be obliged to supply for FRIACO services, if requested. 4.32 The initial ceiling proposed by Oftel to apply from 1 February 2001 is 13,500 tandem switch 2Mbit/s ports. This figure was derived from the forecasts of available tandem switch capacity and demand provided by BT, which are shown in the graph at Annex F to the Panel’s Report. Prior to the final Direction, the figure will be reviewed as part of the detailed analysis outlined in paragraph 4.7. 4.33 The ceiling for FRIACO is appropriately described in terms of switch ports, because this is the unit in which single tandem FRIACO will be purchased. The port capacity for FRIACO services would be available on a ‘first-come-first-served’ basis. Since FRIACO is paid for by an up-front annual fixed charge per 2Mbit/s of FRIACO port capacity, this basis of allocation should not lead to an incentive on OLOs to order an excessive number of FRIACO ports. 4.34 It is proposed that the initial ceilings would be reviewed prior to 30 April 2001. Oftel proposes that there should be a three month period to allow time for the traffic re-arrangement and associated reconfiguration of both BT’s and OLOs’ networks to take place. Therefore, the initial ceiling would not reflect headroom created by traffic re-arrangement. But the ceiling to apply from 1 May 2001 would reflect such headroom and, as appropriate, additional headroom capable of being created. Therefore, it is expected that the ceiling to apply from 1 May 2001 will be significantly higher than the initial ceiling. 4.35 Thereafter, the ceiling could be reviewed periodically in the light of changing circumstances, such as deviations from the forecasts of the demand for tandem switch capacity. To prevent excessive disruption to traffic planning, the reviews should be no more frequently than once every three months. The ceiling arrangements would cease to apply from the start of Stage 2. 4.36 The discussion in this Chapter relates to the creation of headroom in relation to traffic of OLOs who will be purchasing single tandem FRIACO. Oftel has been considering a possible way forward to create additional headroom in Stage 1, beyond that discussed in this Chapter. This would include consideration of traffic re-arrangement in respect of OLOs who will not be purchasing single tandem FRIACO. Therefore, although strictly outside the remit of the request for a determination made by MCIW, Oftel is seeking views in respect of the possibility of further expanding available tandem capacity. Oftel’s views on this matter are set out in detail in Chapter 5 of this document. Stage 2: ISP tandem switch overlay network (from 1 February 2002) 4.37 Oftel believes that BT can expand its tandem switch capacity in the medium term. Indeed, the Panel has identified one way in which this could be achieved, namely through the creation of an ISP tandem switch overlay network. However, the Panel has advised that the creation of such an overlay network requires time to be put in place. The creation of the headroom, using traffic re-arrangement to DLE interconnection as discussed above, should allow BT to provide tandem switch interconnection by using its existing tandem switch network. This will provide sufficient time for an ISP tandem overlay network to be introduced. 4.38 Oftel believes that BT should be able to put in place ISP tandem switches within 12 months from the date of the final Direction. Therefore, Stage 2 will commence on 1 February 2002. Oftel believes that, from the start of Stage 2, BT should be obliged to provide tandem switch capacity for single tandem FRIACO, without a ceiling and without traffic re-arrangement requirements. Prices and Stranded Assets 4.39 During Stage 2, two sets of charges would be relevant. First, the standard pence-per-minute charges for metered single tandem conveyance and the charge per 2Mbit/s port for single tandem FRIACO, as determined under the NCC. Second, there would be an additional charge relating to the provision of new tandem switch capacity, as the need arises. The nature of this type of charge is discussed in more detail in Annex 3 of this document. 4.40 It is for BT to determine how it fulfils the obligation in Stage 2. In order to do so, BT may need to make new investments. It has been suggested that new investment in tandem switch capacity could give rise to BT being left with ‘stranded assets’. In other words, having been required to make the investments, BT might find itself unable to recover the costs of those investments. This issue arises because IP networks appear to be much more efficient than circuit-switched networks for the conveyance of data traffic. Therefore, there would be a possibility that interconnecting OLOs would use the new tandem switches in the short term, but would migrate to other arrangements, such as DLE interconnection or IP-based interconnection solutions in the medium and longer term. The problem would be that this could leave BT with an unrecoverable investment in new tandem switches. 4.41 Oftel believes that it would not be appropriate for BT to bear all of the risk that the costs of its investments might become unrecoverable. However, the possible risk of stranded assets does not mean that interconnection at tandem switch premises should not be provided. Rather, it has important implications for risk sharing and pricing, to avoid exposing BT to an unreasonable risk of failing to recover the costs of its investment in new tandem switches. 4.42 If the stranded asset issue were to arise, Oftel considers that it would not be appropriate for BT to be required to invest in new tandem switches under terms where BT would bear all the risk of the switches becoming obsolete before costs were recovered. It could be argued that such a requirement would not only be unfair on BT, but would also be inefficient by sending incorrect pricing signals to OLOs to feed into their investment decisions. A more reasonable approach might be that that OLOs should bear at least a proportion of the risk. There are a number of ways in which this could be achieved. The issue of stranded assets and initial suggestions on how any problems might be overcome are discussed in Annex 3 of this document. 4.43 The proposed process for setting charges in circumstances of stranded assets would be as follows. The initial trigger would be when BT reasonably believes that it would be required to incur significant capital expenditure, in order to fulfil its Stage 2 obligation to provide single tandem FRIACO without a ceiling. The proposed threshold investment level for the trigger is £2,500,000, which is a provisional estimate of the cost of a new tandem switch and associated equipment. The investment decision would need to be made well in advance of the commencement of Stage 2 (that is, 1 February 2002), to allow sufficient time for the new tandem switch capacity to be put in place. It would be based on traffic forecasts, including those provided by OLOs. BT would then enter into negotiations with OLOs about appropriate charges to recover its investment costs. After 1 May 2001, in the event of a failure to agree, either party could refer the dispute to Oftel for a determination, or Oftel could intervene on its own initiative. 4.44 Oftel proposes to set up an industry group to discuss stranded asset issues in this context. This would serve either to facilitate the negotiations between BT and OLOs, or, if required, to assist Oftel in making a determination without undue delay. 4.45 The problems associated with stranded assets, as described above, could be avoided, or at least significantly reduced, if BT were to choose to meet its obligations by re-deploying its old switches to create an ISP tandem switch overlay network. However, BT argues that it has no old DMSUs capable of being re-deployed. BT intends to retain DMSUs in service at all NGS sites to provide triggering to the IN at the tandem layer until at least mid 2002. BT has stated that at 37 of the current 57 NGS sites, the retained DMSU is fully used for traffic growth in parallel with the NGS. At six of the initial NGS cut-over sites, the retained DMSU is also being used as an STP node for OLO DLE groomed SurfTime and FRIACO traffic. Stage 3: Interconnection at IP level 4.46 Oftel believes that, in the longer term, OLOs will not want to incur the costs of single tandem conveyance on a circuit-switched network. This view is based on the Panel’s advice as set out in the Report. Therefore, Oftel considers that, in the longer term, OLOs are likely to favour two interconnection options. First, OLOs may choose to build out to more DLEs and use their own IP networks to convey traffic. This option is currently available and will continue to be available to OLOs. 4.47 However, it is recognised that it may never be efficient or cost-effective to expect all OLOs to build out to all DLEs, even in the longer term. Therefore, it is likely that OLOs will require a second option of interconnection at the tandem switch premises. The Panel advises that interconnection at IP level is the "best longer-term solution" from a technical point of view (see paragraph 4.10 of the Report). This would enable the traffic originating on BT’s access lines to be groomed off the PSTN by BT onto an IP network at the DLEs. From an OLO’s point of view, with a suitable IP interconnection product, the traffic concentration function, which is currently performed by DMSUs or in Stage 2 by the ISP tandem switches, would be achieved on BT’s IP network. 4.48 As discussed above, the Panel believes that, although broadband access technologies will continue to grow, it is likely that access to the Internet over the PSTN will remain an important feature for the foreseeable future. As the Panel advised, the expected future rates of growth in Internet traffic will place a strain on PSTN tandem switch premises interconnection and could overwhelm the capability of the tandem switch network. 4.49 The Panel expressed the view that IP networks are more efficient than circuit-switched networks in carrying Internet traffic. Oftel agrees that these characteristics make it important and desirable for interconnection to be available at IP level for traffic originating on the PSTN and, in particular, Internet and other data traffic. Therefore, in Oftel’s view, work should commence now to widen the range of interconnection products available to include IP interconnection for Internet access originating on the PSTN. This raises both technical and regulatory issues. Technical Arrangements 4.50 By its nature, an IP interconnection product for traffic originating on BT’s PSTN lines requires the PSTN to IP conversion to be carried out by BT on its modem banks. Traditionally, ISPs have had control over the log-in process, the type of modems used and the contention ratios offered. The log-in process also provides valuable statistics about the performance and behaviour of individual customers. The loss of such features would significantly reduce the attractiveness of IP interconnection, despite its obvious cost advantages. 4.51 Oftel proposes to establish an industry group, chaired by Oftel, to develop the requirements for IP interconnection. Once an outline service description is available, it may be necessary for the Network Interoperability Consultative Committee (‘NICC’) to produce appropriate technical specifications. The NICC is already addressing some aspects of IP interconnection. 4.52 Some OLOs have suggested that, instead of IP interconnection, they might prefer to co-locate their modem banks in BT’s tandem switch premises. This would involve OLOs building out their own IP networks, rather than buying IP interconnection from BT. The Panel noted (see paragraph 6.6 of the Report) that co-location of modem banks does not have the same technical imperative as co-location in the context of ‘local loop unbundling’ and DSL services. Oftel suggests that the industry group, proposed above, could usefully discuss the possibility of co-location of OLOs’ modems, to explore the issues and assess its reasonableness. Market Analysis 4.53 Oftel believes that the nature of any regulation will need to be based on a market analysis. Since regulatory intervention should be proportionate, the likely persistence of any market power could also be an important factor in deciding on the appropriate extent of any regulation. 4.54 One view of the relevant market might be transport of traffic between local switches and the tandem switch premises. Such transport services could be provided using different technologies and types of network. Under this technologically neutral view, single tandem conveyance over the PSTN and IP transport would be close substitutes, so that IP transport would be viewed as a lower cost technology to deliver essentially the same service. 4.55 Oftel considers that BT currently has, and is expected to retain, market power in conveyance over the PSTN up to and including the tandem switch. Therefore, if the relevant market were taken to be a single market for PSTN and IP transport, BT would be considered currently to have market power in data transport between local and tandem switches, irrespective of the technology it used to provide this service. However, the economics of IP networks may be materially different from PSTN technology. It could be that, in the medium to long term, it will become economic for OLOs to build out their IP networks down to at least the local switch level. If so, this could undermine BT’s position of market power in the relevant market. 4.56 However, another possibility is that it would be appropriate to define separate markets for PSTN and IP transport. This might be because the two types of service are considered to have quite different characteristics and so are not close substitutes. Applying this market definition, it would be less clear that BT should be considered to have market power. On the one hand, IP transport could be viewed as a relatively new commercial service, in which it might be premature to identify BT or any OLO as being in a position of market power. On the other hand, it might be that BT’s historical position gives it special advantages in building networks, including IP networks. 4.57 BT can use its existing, ubiquitous network infrastructure, such as switch premises, duct and fibre, and can benefit from its relative density of traffic. Such advantages might extend to IP networks as well as the PSTN. Therefore, it could be argued that BT currently has market power. But, as discussed above, looking forward, it is possible that the economics of IP networks might reduce the likelihood of the emergence or persistence of market power. 4.58 In summary, the two leading alternatives for the definition of the relevant market are a single market, encompassing both PSTN and IP transport of data traffic between the local and tandem switch premises, and a separate market for IP transport. Under either approach, BT might be considered currently to have market power, but the existence of market power would be more clear cut under the single market view. In both cases, it is possible that any initial market power could be eroded over time, if the nature of IP networks makes it economical for competitors to build out to local switches. Regulation 4.59 In the context of IP interconnection, there could be concerns that BT would be able to leverage its market power and thereby distort competition in related markets. For example, a vertically integrated supplier with market power might distort downstream competition among ISPs, by discriminating between its own and other ISP businesses in respect of price. 4.60 There are also a number of non-price features which could be leveraged and thereby distort downstream competition. For example, the operation of the physical log-in process and the customer data it generates. Oftel acknowledges the concerns expressed by the Panel in its Report in respect of the control over this log-in process that would be held by the operator providing the ingress. Oftel notes that the process shows many similarities to Conditional Access in digital television and Access Control Services in other interactive content services. 4.61 Whilst not presuming that all aspects of the Access Control Services regulatory framework would be appropriate for non-price features of IP interconnection, Oftel considers that it could provide a helpful model. Therefore, Oftel seeks views, on whether a regulatory framework similar to that for Access Control Services, should be adopted in relation to non-price features of IP interconnection. A description of the regulatory framework for Access Control Services is at Annex 4 of this document. 4.62 In this document, Oftel is not proposing any requirements on BT in respect of IP interconnection. However, Oftel proposes to set up industry groups, which it would chair, to take forward the regulatory and technical issues associated with IP interconnection.
Further expansion of available tandem capacity 5.1 In Chapter 4 it was suggested that OLOs requesting single tandem FRIACO might have to re-arrange traffic to interconnection with DLE switches which were co-located with tandem switches at which they had PoCs (and possibly also with non-co-located DLEs). Oftel suggested that OLOs could take traffic off the tandem switches in the order of the inverse of the value derived from that traffic’s use of the tandem layer; in other words, traffic whose use of the tandem layer gave least value would be taken off first. Traffic would be taken off the tandem layer until enough headroom had been created to provide for the supply and expected growth of single tandem FRIACO. In this Chapter, Oftel explores the possibility of creating more headroom capacity at tandem switches by extending the traffic re-arrangement requirements to OLOs not purchasing single tandem FRIACO. 5.2 The arrangement set out in Chapter 4, by which OLOs purchasing single tandem FRIACO will migrate traffic to DLEs, will free up tandem switch capacity in the short term. This will create headroom for the estimated growth of unmetered interconnection traffic above the proposed initial ceiling of 13,500 tandem switch 2Mbit/s ports which will apply until 30 April 2001. The availability of this additional headroom will be reflected in the increased ceiling for the period beyond May 2001. 5.3 However, it is expected that demand for interconnection at the tandem switch premises, for both metered and unmetered traffic, will continue to increase, perhaps to a level above the ceiling set for the period after 30 April 2001 under the arrangements discussed in Chapter 4. Oftel believes that, if possible, arrangements should be made so that the extra demand for tandem switch capacity can be met. In order to achieve this, it might be necessary to free up more headroom from the existing tandem switch capacity than can be achieved by the re-arrangement of the traffic of OLOs purchasing single tandem FRIACO. 5.4 The simplest way of freeing up additional capacity at the tandem switch premises would be for all operators, including those not requesting single tandem FRIACO interconnection, to migrate all of their traffic off the tandem switch network onto DLEs co-located with DMSUs, where they have PoCs at these tandem switch premises. This would have the potential to free up a considerably greater amount of tandem switch capacity, and should allow for the conveyance of the rising volumes of metered traffic along with an increase in the ceiling for single tandem FRIACO. This would address the issues identified in Chapter 4. 5.5 In order for such migration to the co-located DLEs to take place, it would be necessary for BT and interconnecting OLOs to agree on plans for such migration, or for Oftel to determine that it was reasonable for BT to require OLOs to migrate traffic in this way. Oftel believes that such a requirement might be reasonable. 5.6 It is quite possible that the headroom created by traffic re-arrangement of OLOs who are expected to purchase single tandem FRIACO will not be sufficient to meet estimated demand for single tandem FRIACO. This suggests that it would be reasonable to take steps further to increase the available headroom. Therefore, Oftel believes that it might be appropriate to require other OLOs also to migrate their traffic to DLEs when they have a PoC at the relevant tandem switch premises. 5.7 It is Oftel’s view that the appropriate charging arrangements for such interconnection would be equivalent to those set out in paragraphs 4.16 to 4.21. Under these arrangements, BT would continue to charge OLOs single tandem rates, but OLOs’ traffic would physically be routed via DLEs and IBLs, STP links, IECs or other links to their switch. In addition, OLOs would pay BT charges for the number of IBLs, STP links, IECs or other links as if they had continued to interconnect at the tandem switch. The IBLs, STP links, IECs or other links needed to re-arrange the traffic to DLE interconnection would be provided by BT without specific charges. But the costs incurred would be forecast and included in the cost base when the next charges are set under the NCC in January 2001. 5.8 This would again be a form of virtual tandem switch interconnection which, from an OLO’s point of view, would be largely equivalent to tandem switch interconnection. The main difference from physical tandem switch interconnection would be that the OLOs might have to use a larger number of ports on their own switches, because their port usage rate might be lower, and manage a larger number of interconnection routes. This would happen because the traffic arising from the co-located DLEs would not be concentrated through the tandem switch. Such concentration of traffic allows higher utilisation of circuits and ports. 5.9 While this might impose some additional costs on OLOs, Oftel believes that these costs would be justified in the interests of freeing up additional tandem switch capacity. Since the physical expansion of such capacity is not possible in Stage 1, and since the forecast demand for tandem switch interconnection capacity is likely to exceed the available supply by a significant amount, some additional costs of freeing up extra capacity are likely to be justified, in the sense that the value of the capacity freed up would exceed the costs of freeing it up. The most efficient solution for all users, and ultimately therefore for customers, is for some small extra costs of lower utilisation to be incurred. 5.10 This approach would fit with both Option 1 and Option 2 as set out in Chapter 4. However, it would fit particularly well with Option 2, whereby BT would have the right to require OLOs to migrate their single tandem traffic onto co-located DLEs. Under this approach, BT would have the incentive to provide tandem switch capacity in the most efficient (that is, cost minimising) way, since its revenues from providing this capacity would be fixed. An extension of Option 2 would give BT the choice of expanding capacity in this way as an alternative to building an ISP tandem overlay network (at least until all available capacity from traffic rearrangement had been released). It is possible that traffic migration of this kind would be a more efficient solution to the problem of a shortage of tandem switch capacity than building an ISP tandem switch overlay network. Therefore, it might be appropriate to allow BT to choose how to provide the capacity required. 5.11 Oftel would welcome comments on the proposal that all OLOs, not just those purchasing single tandem FRIACO, should be required to re-arrange traffic to DLE interconnection in order to free up additional tandem switch capacity. Oftel would also welcome comments on the possibility of extending such traffic re-arrangements into Stage 2 (that is, beyond 31 January 2002) to allow BT to choose between this option and investment in an ISP tandem switch overlay network.
Other options Option B Carriage of dial-up Internet access on circuit-switched narrowband network with interconnection at DMSU switch 6.1 The Panel concluded that capacity on BT’s existing tandem switch network will exhaust before a technically feasible long term solution can be implemented. At first sight, the obvious solution would be for BT to take immediate steps to further expand the network by adding more new switches to its existing NGS replacement programme. The Panel considered this, but concluded that adding to the network will not prevent the forecast growth in traffic levels from exceeding its maximum capacity. It would not be possible to extend the tandem switch network due to the "non-scalability" of the meshed network. There is a factor common to fully meshed networks whereby traffic above around 80 nodes total capacity tails off to a peak beyond which it then starts to reduce. This "non-scalability" is illustrated at Annex B to the Panel’s Report. 6.2 The Panel also concluded that, if the growth in traditional NTS IP traffic, as witnessed since 1997, continues, the total traffic (that is, IP and voice) in the tandem network will double in around two years (this estimate does not take account of further increases in demand created by unmetered Internet tariffs). BT’s existing flat-meshed network simply cannot be scaled to carry that level of traffic without some additional means of absorbing that growth. The ultimate goal has to be the establishment of a wholesale IP interconnection solution for Internet traffic. Until this can be achieved, the only option is to remove as much traffic as possible from the tandem switch network and to provide extra capacity by, for example, an ISP tandem switch overlay network as proposed in Chapter 4 above. 6.3 Therefore, at this stage, Oftel believes that Option B is not an obligation that should reasonably be required of BT. Option C Carriage of dial-up Internet access on circuit-switched narrowband network with interconnection at DLE level 6.4 Option C relies on switching all unmetered traffic via DLEs and extending the capacity to tandem switch premises, but not through a tandem switch, such as by using IECs. 6.5 Oftel notes that this option of interconnection at the DLE is presently already available and will remain available during the three stages described above. Oftel is also conscious that some OLOs have already elected to build out to a substantial number of DLEs, including non-co-located DLEs, either by extending their own networks or by using IECs. Other OLOs may choose to do so in the future. The longer term cost benefits of DLE handover are clear, but the costs of getting to DLEs may prove onerous for smaller OLOs who traditionally have relied on a few PoCs at tandem switches to achieve national coverage. Indeed, many OLOs have not found using IECs to be an attractive option because it creates inefficiencies at low volume DLEs. In addition, for OLOs wishing to offer nation-wide services, it creates significant costs as they need to manage a large number of interconnect routes. 6.6 As set out above, Oftel does not consider it reasonable for Option C to be the only form of interconnection available to OLOs and to expect every OLO to build out its network to every DLE. That is not to say that Oftel wishes, in any way, to discourage DLE interconnection. Indeed, Oftel recognises that this form of interconnection is reasonably required and should continue to be provided by BT. It is clear that the greater the volumes of traffic that can be removed from the tandem switch network layer, the longer the network can support residual traffic carried to tandem switch PoCs. In turn, this buys more time to establish wholesale IP interconnection for Internet traffic. However, Oftel believes that BT should provide single tandem conveyance if it is technically feasible to do so. 6.7 Oftel believes that Option C should remain part of the FRIACO product set, but that it is not sufficient, on its own, to meet demands.
Double tandem conveyance 7.1 Oftel believes there is a need to clarify the question of how many tandem switching stages can be used for the conveyance of FRIACO calls over the PSTN to OLOs’ PoCs. 7.2 In order to minimise the risks to the PSTN, the objective must be for the residual volumes of unmetered traffic, not groomed at DLEs, to be handed over at the first tandem switching stage and not to allow it to cross the DMSU mesh. Therefore, OLOs will be required to interconnect at sufficient tandem switches to achieve single tandem routing of FRIACO calls to their PoCs. For OLOs offering services in limited geographic areas this may require only limited changes to their existing interconnection arrangements. OLOs requiring national coverage may need to discuss with BT the number and location of tandem switches they need to reach in order to achieve single tandem interconnection. 7.3 In short, Oftel is not proposing a ‘double tandem FRIACO’ product option.
Numb | ||||||||||||||||||||
