THUS plc
1-2 Berkeley Square T 0141 567 1234 99 Berkeley Street F 0141 566 3010 Glasgow G3 7HR www.thus.net
THUS plc
Registered Office: 1/2 Berkeley Square
99 Berkeley Street Glasgow G3 7HR
Next Generation Networks
Future arrangements for access and interconnection
Response to Ofcom Consultation from THUS plc
13 January 2005
Introduction
THUS is pleased to respond to Ofcom’s consultation on ‘Next Generation Networks – Future arrangements for access and interconnection’. Our response covers various issues of particular concern to THUS. We have not responded directly to Ofcom’s questions, but many of the questions are addressed by the points raised in this response. This response represents a snapshot of current thinking, which is likely to evolve as more information becomes available and our understanding of the different options improves.
THUS is already implementing its own next generation network and welcomes BT’s plans to upgrade its own network. This should help raise awareness and market demand for the new types of service that can be offered over NGNs, and with suitable interconnection arrangements will allow interconnecting operators to offer end-to-end services across networks. It also presents a unique opportunity to design networks and systems from the outset in a pro-competitive way that respects the principle of equivalence of input set out in Ofcom’s strategic review.
However it must be recognised that BT is investing in its ‘21st century network’ for the benefit of its
shareholders, to reduce costs and gain competitive advantage; not for the benefit of UK plc. Although there are opportunities, BT’s plans also represent a grave threat to alternative operators, through the possibility that:
• substantial unrecoverable costs and disruption will be imposed on BT’s interconnect partners as a result of its decision to upgrade its network; the impact will be particularly damaging if BT presses ahead with migrations, exchange closures etc to an aggressive timescale which leaves insufficient time for thorough testing and planning;
• many of the advanced services and capabilities which can be offered by virtue of the NGN will be available to BT Retail and other retailers of end-to-end services long before next generation interconnects have been defined and implemented, thereby giving BT and its resellers first mover advantage what will be key business markets;
• despite ‘in principle’ commitments to the principle of equivalence, the opportunity to engineer the network in a pro-competitive manner is thwarted by a superficial consultation process with industry which fails to share information or allow debate of previously taken design decisions;
• next generation networks will allow BT to exploit new economic bottlenecks and sources of dominance; for example BT lists digital rights management as one of the common
capabilities to be offered by 21CN; we believe DRM is a service that should be capable of being offered commercially by any provider and are disturbed that BT should be attempting to bundle this as part of the core network functionality.
We are pleased that Ofcom has welcomed many of these issues in its consultation document. However, we believe it is vital that Ofcom becomes proactively engaged in the planning and consultation process from an early stage. We agree that in general Ofcom should allow industry negotiation to run its course and intervene only when disputes arise. But in the present case the decisions are so critical and the timescales so short that Ofcom cannot afford a laissez faire approach. It is vital that Ofcom intervenes at an early stage to set the parameters of the negotiations, and to signal clear sanctions and remedies against BT should BT force through decisions or changes to its own timetable without adequately meeting the needs of altnets.
Physical access
We agree with Ofcom that access and interconnection are required at both the MSAN and Metro node levels. We also agree that the amount of competitive pressure at MSAN sites will vary geographically; it is clear that a cluster of co-located MSANs in a prosperous area will be more attractive than a single MSAN in a depressed one.
However, we do not believe that it is practical to draw a line in advance between these two classes. The attractiveness of access to a site will depend on a number of factors that will vary between operators. For example, a site that might otherwise be ignored could be attractive to a regional operator wishing to "complete the set" so that they can claim 100% coverage. Nor is it reasonable for Ofcom to attempt to second-guess operators in this matter. Therefore access and interconnection should initially be available, in principle, at all MSANs.
Provided that suitable arrangements are made to deal with sunk assets, we accept that it would be reasonable to exclude interconnection at MSANs using legacy protocols where this would require the addition of gateways purely for this purpose.
Service layer access
We do not believe that Ofcom’s first principle can reasonably be applied to service layer access at the MSAN. As the consultation points out, this would imply that physical unbundling should be the long-term goal of regulation. This would mean that there would be no effective competition at those MSANs where there was insufficient demand for unbundled service. Furthermore, if only one or two operators deem a site worth investing in, it would potentially leave those sites with an oligopoly market in downstream services, rather than a truly competitive one.
There are three types of access at the MSAN site that need to be considered: 1. Access on the customer side of the MSAN (e.g. unbundled loops).
2. Access within the MSAN (e.g. wholesale line rental).
3. Access at the network interface to the MSAN (e.g. bitstream).
These three do not form a simple progression where it can be said that one can be built on top of another. Rather, we believe that they form separate markets with different competition issues. We believe there will be a need for continuing regulation to protect competition at each of these levels, though not necessarily at every geographic location. This is not to denigrate Local Loop
Unbundling as a solution, merely to note that it is not a regulatory panacea.
Competition in the UK telephony market happens at both the service and transmission layers. For this competition to be effective – outside the relatively small number of players who can take part in LLU – there need to be competitive mechanisms applicable at each layer. These correspond, at least approximately, to the second and third situations listed above.
First consider transmission. There is already a vibrant market in this area based on
interconnection at both DLE and Tandem. Quite apart from the transitional issues discussed later, the existence of this market would be jeopardised by being restricted to the Metro node layer, because a significant part of the transmission of any call would be on the monopoly MSAN-Metro links, reducing the available margins that drive such a market. Furthermore, having to rely on the BT link would restrict the options for providing services with different levels of quality, something that matters even more in the new data markets being opened up than it does with voice. More
generally, MSAN interconnection at the bitstream level will be a vital part of the provision of new services; it should not be seen as merely a way to allow operators to compete with BT on price. Therefore we strongly urge Ofcom to ensure that a regulated transmission interconnection product is available at the MSAN.
We do not believe that this interconnection product need be particularly expensive or difficult to provide. It basically involves adding an IP router – a relatively inexpensive and reliable device – at each MSAN site where interconnection is desired. Standard IP routeing mechanisms will then ensure that traffic flows over the appropriate links. We also note that this interconnection is only important for the “media” flows on each connection; the signalling data is small enough that we would not expect efforts to be made to divert it from the “natural” route to the call servers.
BT has, in recent days, suggested that an ATM switching layer could be provided at low cost but an IP one would be extremely expensive (doubling the cost of a typical MSAN site). While we welcome this step forward, we remain sceptical that the proposed design is optimal. We look forward to taking part in further discussions.
Moving now to the service layer, it is important to note that competition at this layer is largely independent of that at the transmission layer. In particular, the provision of service “interconnect” products at the MSAN does not prevent these services being backhauled over BT lines and, conversely, the provision of transmission interconnection does not require new competitive products in the MSAN.
Having said this, we only see a need in the short term for the equivalent of Wholesale Line Rental, with the renting operator providing the call server for all calls from and to a particular line. It would then be up to the operator to provide, or rent, transmission for the calls. There appears to be no effective difference between this and Carrier Pre-Select, other than that CPS involves relaying signalling traffic (essentially unaltered) through BT’s call server while WLR does not, and so there is no need for a separate regulated CPS. However, Ofcom may wish to consider retaining a separate Indirect Access product, as this would allow for competition on a per-call basis. In the longer term, we believe that competition will require some kind of deeper regulated access to the MSAN software, possibly up to and including the ability of an operator to install their own
replacement for the H.248 or SIP client for those lines they are renting through WLR. In summary, and to answer question 6 directly, we see the need for the following regulated interconnection products at the MSAN:
• Local Loop Unbundling
• interconnection for transmission links
• call routeing on to such links by the BT call servers
• Wholesale Line Rental with all traffic using the renter’s call server • Indirect Access if per-call competition is seen as desirable
• (future) control or replacement of the H.248 or SIP client in the MSAN.
We believe that paragraphs 3.53 and 3.54 of the consultation document badly misrepresent the situation with IP routeing. In the switched telephony world it is natural for all traffic from and to a line to follow the same path through the network. This is not the case with IP. On the contrary, it is normal and expected for different traffic to follow different paths and, in particular, for the outward and return traffic of a single communication to be routed differently.
Paragraph 3.54 suggests that it will be cheaper to route all traffic from a given line card the same way than to route based on final network destination. This is not the case for IP either. A router makes a decision for each packet based on the packet header; indeed, it is far more usual to route
based on the destination of a packet than on its source! If the call server assigns an appropriate destination address for the media connection of a call, it will automatically follow the correct path. Thus “Soft LLU” is no better – and indeed is probably worse – than “bitstream interconnection”. The assumption in the document – which seems also to be held by BT – is that using ATM or MPLS virtual circuits direct from the line card is enormously cheaper than using IP routeing. We cannot identify the reasoning behind this assumption – the relevant equipment is similar in cost – and therefore we consider the “layer 2 v layer 3 v voice interconnect” choice of question 15 to be largely false.
We would also note that there are lessons to be learned from experience of DataStream and IPStream. OLOs were effectively presented with a fait accompli as regards the design of DataStream and, after years of fixes, it is still barely fit for purpose. It is vital that BT consults openly on the design of the successors to these products and gives the industry much greater visibility of the relationship between its own upstream and downstream products.
In respect of question 16, we do not wish to rule out service-specific interconnect, though as described above we consider it to be largely separate from the IP routeing question. We do, however, accept that TDM interconnect at the MSAN is unnecessarily if proper control of the IP connection is available to operators.
Finally, we are very concerned at reports that BT's specifications for MSANs make no allowance for any kind of interconnection at all. This would appear to be an attempt to impose a position on MSAN interconnect without any form of meaningful consultation. If so, we would expect the costs of rectification to be met entirely by BT and specifically excluded from any regulated charges.
Equality of Access and Regulatory Withdrawal
We agree with Ofcom that equivalence of input, rather than equivalence of outcome, should be adopted wherever possible and, in particular, for all 21CN products. However, we disagree that it should be used only sparingly with legacy services - on the contrary, equivalence of outcome should only be used where there is an exceptionally strong justification for allowing discrimination between BTR and other operators. While equivalence of input requires more effort to put into place, it will produce much less need for intervention later; on the contrary, equivalence of outcome involves the ongoing burden in policing KPIs if it is to be effective. We do accept that, where a product is unlikely to last more than a year or two, equivalence of outcome might be more cost-effective, but we expect most legacy products to last at least long enough to make this
dubious.
We do not foresee any area related to 21CN where regulatory withdrawal would be sensible in the short or medium term. We do agree with Ofcom’s proposals to review the remaining wholesale voice regulated products once migration to 21CN is substantially complete, but we expect that this will be more likely to be 2010-11 than 2008.
We agree with the suggestion in paragraphs 3.32 and 3.33 that initially BT should be required to make all options available at all locations where there is not yet evidence of effective competition. However, we believe that triggering the first phase of regulatory withdrawal on the basis of LLU orders is unnecessarily aggressive: it serves no additional purpose and risks damaging existing competition based on IPStream and DataStream in the transitional period before adequate
wholesale alternatives become available. The fact that one or two operators have started ordering unbundled loops in an exchange is no guarantee that adequate wholesale alternatives will emerge on a competitive basis. If most of the local loop unbundlers decide to remain vertically integrated
(which is not unlikely, given experience of the mobile market) there will be a significant number of exchanges where resellers will be forced to continue to rely on BT Wholesale, who will therefore continue to enjoy market power at the bitstream level. Furthermore, because many of the costs are fixed, the market may only be able to support two or three LLU operators; "light touch regulation" should not simply result in an oligopoly of the early arrivals.
Network Charge Controls
The current Network Charge control regime is due to expire in October of this year and many of the key network, interconnection, and product decisions surrounding the move to 21CN
architecture remain undecided. This has led THUS and others to argue in the past for a shorter charge control period, giving Ofcom and the industry more time to appreciate the effects of 21CN. However, Ofcom has indicated that it does not share this view, at least partly because data on 21CN deployment and traffic volumes may not be sufficiently detailed to facilitate an IP-based control much before the end of the next 4-year period.
THUS recognises the stability and certainty benefits that a further four year control period would provide. However, Ofcom must ensure that any new control is not just a roll-over of the current one, but is based on a “clean sheet” analysis of BT’s costs. The new control should be technology- neutral, smoothing out any extremes in the costs resulting from the migration of traffic from PSTN to 21CN (e.g. high start-up costs for 21CN and loss of economies of scale on both networks because of lowered traffic volumes). The new control must also preserve the current Element Based Charging structure and differentials so as to provide the industry with continuity until the move to 21CN-based charging.
Inter-Metro Conveyance
In principle, inter-Metro node conveyance is a natural progression from inter-tandem conveyance and should be fully competitive. However, there are two major factors militating against this: 1. Many of the Metro node locations are not the same as any existing point of interconnection.
Therefore there will not be any existing paths for competitive traffic flows.
2. MSANs will be single-parented, unlike the present arrangement where DLEs are connected to several tandem exchanges. Thus "single metro" connectivity will involve approximately three times as many points of interconnection as "single tandem".
If 21CN had been introduced 7 or 8 years ago, it is likely that every Metro node would have had physical connections dug by several operators. But the economic conditions are much different today, and we could not be confident that any operator is willing to extend their network to Aberystwyth, Elgin, or Truro. We therefore suggest that it will be necessary to provide some kind of regulated inter-Metro transit for the forseeable future. We do not yet have a firm idea of the best form of this facility. However, the organisation of traffic flows in a packet-switched network suggest to us that the situation should not be treated as separate MSAN-to-parent-Metro and inter-Metro products, but rather as a MSAN-to-remote-Metro product. That is, whatever regulated products are made available via interconnect at the parent Metro node should also be available with the point of interconnect being at any other Metro node.
Network Hooks
We feel that there is a line to be drawn between “intelligence capabilities” and “network hooks”. The latter tend to be a matter of interfaces and are hard to differentiate from network connectivity and protocol conformance. The former, on the other hand, are a further layer of abstraction away from the network, and tend to be more related to control and information.
We consider Quality of Service and Call Prioritisation as network hooks, and they should be considered as part of the basic interconnection regime. Once the functionality is defined, it will be possible to consider how to overlay intelligence capabilities and what the impacts are on ordering, service, and billing. To this extent, they are no different from any other new interconnect method introduced on the present network, such as DLE interconnection. Because of the switch from circuit to packet-based communication, we consider Quality of Service to be a fundamental pre-requisite for all service development on 21CN.
We consider Authentication, Presence and Location, and Directory and Profile (which we take to be a middleware layer between control and the physical network) to be intelligence capabilities. We would add to these “Status” (real-time information about connections) and “Statistics and Reports” (which would be non-real-time information about usage, performance, and analysis of logs). Intelligence capabilities should not depend on the particular method of network access and, in particular, whether it is at the MDF, the MSAN, or the Metro node.
We are unclear what is meant by “Secure Connections”, and so cannot comment on it.
We do not consider that Digital Rights Management should be part of 21CN functionality; it is a business application and not a network interconnection facility. If it requires features such as authentication, those need to be addressed on their own merits.
In answer to question 22, we would suggest that BT should face specific access obligations at least under the following circumstances:
1. Where making information available to BT Retail could result in service differentiation. Some examples would be:
• services that make use of additional authentication attributes (e.g. parental control); • improved bundling and co-ordination of services;
• improved fault-finding because connection information can be accessed online.
2. Where the development of new services is unnecessarily restricted because BT have chosen not to develop them themselves.
3. Where the lack of intelligence results in operational or development costs to service providers. For example:
• lack of access to authentication could result in legal and regulatory exposure;
• lack of access to real-time information and end-to-end visibility increases support costs. 4. Where the lack of intelligence would prevent equivalent services being provided via LLU and
via BT MSANs. It should not be necessary for services to vary depending on where the customer is connected.
This last case applies equally to question 23: regulation, of whatever form, should ensure that sufficient intelligence is present to allow transparency between equivalent BT and LLU services; for example, presence and location information should be available on equal terms.
We are unable to suggest to what extent commercial negotiation might remove the need for regulation, because it depends almost entirely on the extent to which the strategies of different operators (including BT) align. However, we accept that at least some capabilities (e.g. access to
statistics) may impose a significant overhead on BT and therefore should be optional – operators may wish to negotiate a price but should not be forced to.
In respect of question 25, it is certainly important to consider the provision of deeper hooks to directly access intelligence capabilities. This is not distinct from what has been discussed above, but is simply part of a continuum in the spectrum of required interfaces. API standards will always change over time and the pros and cons of each one depend on the detailed requirements, so it is less important to consider the specifics of the standards than to minimise the number of protocols used, offer stability, and ensure change is introduced in a timely and structured manner.
Next Generation OSS
We have consistently held the view that equivalence of inputs is far preferable to equivalence of outputs in most situations, and this is no exception. Without it, the potential for shared intelligence capabilities is significantly reduced and service providers will typically have little option than to offer non-differentiated services from a narrow BT wholesale portfolio; on the other hand, the benefits of providing it will include optimised intelligence, operational costs and service
differentiation. However, not all areas will require direct access to OSS; in some cases it may be sufficient to provide equality of access to the 21CN middleware and common directories.
The barriers to achieving equivalence of input are more logistical than technical. If BT adopt common third-party systems and industry standards, integration will normally be feasible. The problems are much more likely to involve failures in the following areas:
• providing timely availability of specifications, roadmaps, architecture, and schedules; • keeping to agreed schedules and specifications;
• agreement on universal data structures;
• establishing common infrastructure (e.g. security, directories, access, etc.); • maximizing stability and minimizing change;
• providing readily available test environments;
and, of course, problems induced if BT introduce their own incompatible systems after several service providers have already invested and installed NGNs.
All of these should be surmountable, though BT’s track record suggests that significant improvement is required.
Number Portability
The introduction of centralised call servers, which will make all call routeing decisions, means that it may be time to revisit the whole concept of geographical numbering and number portability. With a traditional telephone network the telephone number is used by various pieces of equipment in the network to route a call. There is therefore a benefit to allocating numbers in blocks that relate to a specific exchange. With 21CN, however, the actual routeing of traffic will be done using the IP addresses of the MSANs, while telephone numbers will only be of interest to the call servers – they will become more like domain names in the Internet environment. At this point, migrating a phone number to a different MSAN should be no more difficult than moving a number between lines on a single MDF (which is already no more than changing an entry in the exchange configuration tables). This therefore means that there can be a more relaxed approach to the geographic significance of numbers.
For example, consider the area covered by a 10km radius around Cambridge. Most of this is in the 01223 dialling code, though parts of the northwest are in 01954, and lines connect to several different concentrators. If a resident moves house within the area, it is a gamble – to them – as to whether they can keep the same number, get a new number with the same code, or get a number with the other code. Once call servers are used to route calls in this area, there is no technical reason why a customer should not be able to retain the same phone number if they want, even though they are now served by a different MSAN or resident in a different dialling code area. Furthermore, allowing this blurring of the dialling code boundaries would improve, not harm, the perceived meaning of the 01223 dialling code – it becomes “Cambridge and surrounding area” rather than “Cambridge and most of the surrounding area”. There are no doubt many other places where this relaxation would be beneficial.
More generally, this move to call servers should mean that there is no longer a need to restrict the use of “out of area” numbers. The recent New Voice Services situation has shown that there is a demand for numbers associated with the location customers wish to be associated with, rather than their actual physical location, and 21CN should be used as a way to make this more available, rather than impose arbitrary restriction.
The same reasoning implies that number portability between operators – at least for those using call servers rather than converting back to TDM – should be equally simple irrespective of geographic location. This should be seen as an opportunity to loosen restrictions on porting.
Transitional Arrangements
We agree with the guiding principles laid out in paragraph 4.5.
Geographic rearrangement will be a major issue. Operators will have designed their networks at least partially based on the interconnection arrangements with BT. This leads to various problems: • Some existing tandem exchanges will not be Metro node locations, and tandem interconnect
links will therefore become "stranded assets".
• Even if BT offer free conveyance from the existing interconnect point to a nearby Metro node, the connection may no longer be appropriately sized for the traffic flows to and from that node. • Until the arrangements for MSAN interconnect are known, it is hard for operators to design a
new interconnect strategy. This makes it difficult to comment on proposals to provide conveyance from existing POIs to Metro nodes.
We are skeptical that BT will be quick to offer sufficient information and appropriate levels of compensation; whether or not the goodwill is present in this instance, their track record is
discouraging. We would expect Ofcom to be ready to bring forward proposals at an early stage if clear progress is not being made.
We expect most operators to migrate to IP in a relatively short timescale. Therefore – provided it is possible to use physical links efficiently, which implies proper control over the IP level and
interconnection at the MSAN – we see the provision of TDM interconnect at a large number of locations as being less important. We would therefore be content with provision at the Metro nodes only. We are unclear what arrangements are being suggested in 5.15 and therefore are reluctant to comment; how would it be determined if an operator was “capable of providing backhaul”? We continue to believe there would be strong benefits to retaining the current EBC regime for interconnection charging for the next control period (to September 2009). As discussed earlier, this
needs to be technology-neutral: we would suggest that a snapshot be taken of an operator’s PSTN interconnection at some point in time, and this be used for charging purposes irrespective of subsequent changes because of 21CN. As BT is requiring operators to change their
interconnection arrangements, they should meet the reasonable costs of this work; this proposal would address the per-call components (though, of course, not the actual costs of physical rearrangement).
Flat Rate Charging
In respect of question 33, we agree that there is scope for new arrangements for voice charging. While it is too early to be sure, we suspect that the majority of network costs in the future are going to be dominated by the fixed infrastructure of the local loop and MSAN. As such, we would expect a major rebalancing of tariffs to take place and unmetered tariffs – at least for calls within the UK – to become much more prevalent. At the wholesale level this would map into capacity-based
charging, with operators being charged on the basis of the maximum number of simultaneous calls being handled or (for data interconnections) the available bandwidth. Call termination fees, in particular, could ultimately become obsolete.
However, caution is needed before assuming that the future should simply use capacity-based charging. Firstly, it is important to consider the NTS regime. This is a micropayment system that the UK should be proud of – despite some issues at the margins, NTS is extremely successful. Ofcom should ensure that it is not damaged by the transition to new charging methodologies. Secondly, it may not be sufficient to rely purely on capacity-based products; the products must be correctly designed for the market:
• Voice products probably need to be priced per-customer, while data products would be per Mb/s.
• It must be possible to grow capacity in relatively small quanta: a 50% or 100% price increase in wholesale cost to cope with a small growth in customer base is not sustainable.
• There should be combination products that allow for both a permanent capacity and for short-term overflow (unlike IPstream, where ISPs have to choose either CBC or UBC); obviously the latter would attract a premium. The overflow arrangement could be capacity based (perhaps charged on the maximum overflow each week or month) or it could be based on actual traffic (call-minutes or megabytes).
Converged Access
While we see benefits in combined line access products as discussed in paragraph 5.9, we believe that the market will continue to distinguish "POTS voice" from "New Voice Services" for some time to come, if only because the former provides reliable 999 service while the latter does not. This leads us to conclude that it should continue to be possible for operators to lease only the narrowband or only the broadband part of a local loop. Furthermore, this applies not only at the LLU layer (which is never going to be a practical approach for much more than 50% of the UK), but more importantly, at the WLR layer (by providing partial access to the MSAN interfaces). We have significant concerns about being restricted to combined line access products. It is quite possible that BT could deploy voice services on the “POTS” part of a line that we were unable to replicate (perhaps because of a lack of end-to-end interoperability, or perhaps for commercial reasons). If so, being limited to combined access products would make it impossible for us to provide broadband (or “NVS”) to customers who wanted to subscribe to BT’s new voice services.
In effect, BT would be able to leverage its dominance in the POTS market to squeeze us out of the broadband market.
Related to this, we see no technical reason why provision of voice and broadband to a specific customer over separate bitstream interconnects should be either difficult or expensive.
The question of converged access is part of a wider issue: the granularity or specificity of regulated facilities. It is important that the choice of such facilities is made in a way that allows operators to design their own end-user products, rather than limiting them to a fixed “menu”. For example, there should be options for various levels of quality of service, including mixed levels over the same connection, rather than BT or Ofcom assuming that there are “right” settings for voice, for consumer broadband, for PPC-replacements, and so on.
Consumer Data Services
We do not feel that it is within Ofcom's remit to determine whether provision of voice-band modems remain efficient just because of changes to the underlying transport. In particular, it should be noted that the equipment used can have a significant effect on the customer experience, and therefore ISPs may wish to continue to provide their own facilities rather than rely on that provided by BT. We would argue that provision of dial-up Internet service can, in general, be left to the open market. However, the continued popularity of SurfTime and of products based on
FRIACO show that there is significant demand at the retail level which can only be satisfied by a wholesale capacity-based product. This does not mean that these products must remain
unchanged, but there must be suitable replacements, even if generic rather than specific to low-bandwidth IP.
Moving to the broadband area, we agree that there is no particular need for a regulated service specific to consumer broadband. We also agree that future services may require some form of managed QoS. In answer to question 57, we do not think that anyone is yet in a position to say whether generic or specific facilities will be needed; it may well be that both will be required as new services are designed.
We are of the belief that both layer 2 (ATM or Ethernet) and layer 3 (IP) interconnection will be required. IP is not well-designed for delivery of layer 2 services, and without a clear arrangement for their reliable delivery over IP a form of layer 2 interconnection will continue to be required. Simple provision of a QoS facility is not sufficient to address this. While there will be technical issues in the conversion from ATM to Ethernet (e.g. virtual circuit numbering and identification) these do not affect this basic requirement.
Finally, we note that all these services (other than those involving interconnection at the physical layer) rely on authentication. Even if other parts of the product space can be left to the market, this is likely to require some form of regulation.
Business Data Services
We do not agree with the claim in paragraph 5.35 that QoS is an important distinction between business and consumer data services. Many businesses are content with 20:1, or even 50:1, contended services based on IPStream while, at the same time and as already noted, consumers will increasingly demand QoS for advanced services like video. Therefore this distinction is likely to disappear with time.
As the consultation points out, the presence of Partial Private Circuits (PPCs) means there is little need for regulation of other products in this arena. Because of this, we are particularly concerned that there are little or no details available on what facilities will replace PPCs in the 21CN.
It would seem obvious that there will be an ongoing need for a partial private circuit product of some kind, available at a range of speeds, and we consider this to be essential. The same issues apply to these services (e.g. the need for QoS control and the desirability of handover at the MSAN) as to voice services.
In answer to questions 61 and 62, there are a number of applications that have stringent restrictions on circuit latency and jitter (examples are those that require ATM CBR at present). These will be poor candidates for migration to emulated circuits. In particular, we are aware that the electricity supply industry has very stringent limits on the acceptable latency on circuits used to control their networks – in particular, to shut down systems – and consequential concerns that IP-based networks cannot meet those limits. Having said that, if the emulation is truly transparent (that is, the emulated circuit meets the actual properties of a native circuit and a customer cannot distinguish the two by testing), there should be no technical issues.
We disagree with the views expressed in paragraph 5.39. We expect Ethernet services to evolve to include multipoint, and in any case disagree with the notion that they do not lend themselves to the PPC model. We do not believe it is desirable to rule out any particular cases at this stage, and therefore all the variants listed in question 65 need to be considered. There is also a requirement for some kind of “protected” or “resilient” interconnect to substitute for APS. At the same time, there will remain a need for PDH/SDH services (or perfect emulations) and these cannot simply be considered “legacy”.
In response to question 70, issues that would need to be resolved include authentication, the availability of layer 2 services, identification of bitstreams (e.g. virtual circuit numbering), and resilient/protected interconnect.
