Top Story

As legal IP telephony spreads out across the continent, internet-based practices may begin to take over from past telephony ways of doing business. Russell Southwood looks forward to how this might affect interconnect issues and the whole structure of the industry.

Two shifts are happening at a global level and the second is a direct consequence of the first:

• The shift to IP-based business practices: Telecom and Internet operators do it differently at a number of levels. Put simply, at an international level telcos charge for “half-circuits” whereas the Internet operators charge for a full circuit. Telcos base their charges on geographic distance whereas distance is much less relevant for Internet operators. Telcos charge for time whereas Internet operators charge for capacity. Traditionally, telcos have predicated their operating assumptions on a centralising technology with them at the centre whereas IP networks facilitate a layered network approach that allows “many-to-many” relationships.

With the increasing uptake of VoIP by traditional voice operators, larger amounts of traffic are based on IP data. Whereas in the past these two different parts of the industry operated in completely different currencies (minutes and bandwidth), they now have a common currency (bandwidth). So whether the telco peers with an ISP to carry VoIP-based traffic on the internet or buys dedicated bandwidth for those calls, it increasingly takes on the business practices of those that carry data.

Like all transitions in industry practice, this will not happen overnight and the picture will look much more confusing than the future that we describe below. Several large telco carriers are both voice operators and ISPs and they will continue to use both sets of industry practices for some time to come.

• The shift from minutes to “free” and bundled voice access: Internet companies sell capacity and have continued to do so in offering new VoIP-based services. Most of the current packages include a monthly payment (usually based on broadband capacity) and allow users to call other subscribers for free.

Both elements allow for unlimited voice use limited only by time itself: the number of hours an individual can talk. “Break-out” calls that go to PSTN numbers are charged for by the minute but this may just be a transitional element of old telco practice.

This shift is happening fastest in traditional voice-telcos that are seeking to leverage the use of their fixed line assets to become the largest broadband players in their market. In this way, these operators feel they can re-invent their central position with IP.

They want to be able to remain centrally positioned by offering the “triple-play” combination of voice, the Internet and “television”. The latter is more probably sport and films either with IP-TV or downloadable material. In the African context, there are currently only four players with this future in mind: Senegal’s Sonatel, Gambia’s Gamtel, Mauritius Telecom and South Africa’s Telkom. Sonatel is trialling IP-TV, Gamtel intends to do so shortly, Mauritius Telecom will launch on 15 March and Telkom is rumoured to be interested in buying a content-based company.

The shift to Internet-based practices will be based on the frustrations and inefficiencies of the existing hybrid practice. Currently if you use one of the major consumer VoIP providers (for example Skype, Vonage or Delta Three), your calls will need to travel across the PSTN at some point, despite the fact that both the VoIP provider and the telco are using IP-based networks to connect the calls. The same situation is mirrored in the corporate world. Two companies may both have installed VoIP systems but in order to ring each other they need to go through the PSTN.

In many cases, these calls that need to go through the PSTN will be subject to a packet-to-circuit-to-packet conversion. This conversion process is inherently inefficient and leads to a reduction in call quality. As the transition to VoIP in the early stages involved only a relatively small amount of traffic, both parties have been willing to put up with these inefficiencies. With greater levels of traffic, VoIP operators inevitably began to ask themselves why should they continue to “subsidise” the analogue sunk costs of the PSTN when it can be done cheaper using direct IP connections?

As a result, third party clearinghouses have been set up in Europe and the USA to exchange VoIP traffic. One of the third-party clearinghouses that set up in June last year in the USA was Stealth Communications’ Voice Peering Fabric. Also US operators Equinix and NeuStar got together to offer a similar service. Equinix is well-known in the Internet sector as the operator of 15 Internet Business Exchange (IBX) data centres.

From the perspective of an African operator, these developments might seem like something that’s ‘a long way away and in another country that’s so different to our own’. But this is to forget the way that the introduction of VoIP caught up with operators on the continent. It encouraged grey market operators that forced down prices and the incumbent telcos were themselves forced to respond by installing IP gateways (usually provided by outside operators) in order to fight “fire-with-fire”.

And where there has been the legalisation of VoIP and the introduction of broadband, telcos have again found themselves dragged forward by the now legal clamour of IP-based operators: the IP-based operation is just so much more efficient and again they have had to find ways of keeping traffic by offering services themselves. And so it will be with carrier VoIP peering and enterprise VoIP peering: unless they seize the initiative, Africa’s incumbent telcos will find themselves following rather than leading these developments, as international developments again set the pace.

Some figures give some idea of what’s on the horizon. At the beginning of 2005, VoIP-peering exchange Voice Peering Fabric set itself an annual run rate of 9 billion minutes for the calendar year of 2005. In the event, it ended November of that year having achieved 18 billion minutes.

Although the majority of minutes are originating within the USA, a survey of Voice Peering Fabric’s members showed that half of all calls peered were international terminations. And this is not solely an American, big market phenomenon. In Australia, VoIP carriers like Unwired, Internode and Pacific Internet connect their networks together again in datacentre specialist Equinix’s facility to avoid the cost of pushing more traffic onto Internet backbones owned by large carriers like Optus and Telstra.

So is this kind of VoIP peering relevant to smaller African carriers? In some ways it is probably more useful than it might be for larger carriers with their own well-developed international relationships. One of Voice Peering Fabric’s more recent members is Telecom New Zealand and through its membership it gets what are claimed to be “quality direct routes with more than 300 interconnect relationships in the United States and around the world.” Africa will increasingly become what Voice Peering Fabric describes as “newly deregulated markets.” VoIP peering is a cost-effective way to extend a carriers’ reach by providing interconnect to small carriers in emerging markets like Africa. And for the small African carriers it might also extends their reach globally.

The economics of IP peering are based on a clear business case. Calls sent via TCP/IP reduce the need for costly PSTN network interconnections and enable free calling between two users. Calls sent by IP that do not go over the Internet also can offer similar if lower savings. Furthermore, as with IXPs, there is no real reason why VoIP peering exchanges need any regulation.

Two things need to happen at a practical level to make these savings an attainable reality:

1. VoIP peering: As described above, there needs to be either mechanisms like peering exchanges or straightforward, one-to-one peering relationships between carriers and indeed between enterprises. ‘Islands’ of VoIP providers will interconnect directly avoiding the TDM-based networks of the PSTN.

2. ENUM or its equivalent: The IP-based call needs to be able to find the exact location where the call can terminate, whether this an existing PSTN number, a VoIP provider allocated number or an IP address. Most operators use a mixture of peering and transit but the smaller operators almost exclusively buy transit, whereas the larger operators rely upon selling transit to make their margins.

VoIP peering differs from ISP peering in a number of respects. It can either happen directly over the public Internet or as in multilateral peering via an exchange of some sort. Also in some versions of VoIP peering operate on the basis of billing for calls per minute over IP: Voice Peering Fabric operates this way which is why its traffic forecasts and traffic carried given above are in minutes. But a non-public WAN or VPN becomes a private voice network and calls are carried at a flat rate. Peering between these types of private voice networks would almost exactly mirror ISP peering arrangements.

Bilateral VoIP peering agreements may or may not include “minutes” as part of the agreement. By contrast multilateral VoIP peering will not include a charge per call or per minute.

So how does this work in practice? Larger Tier 1 ISPs like MCI and AT&T (who also happen to be voice carriers) interconnect without any charges (through a bilateral peering agreement), since each ISP benefits from the more direct linkage. Smaller Tier 2 ISPs must usually pay to interconnect to a Tier 1, since they need the connectivity more than the Tier 1 ISP does.

In this new world African telco incumbents might have a bilateral peering VoIP peering arrangement with a large Tier 2 ISP (like Internet Solutions in South Africa) or an alternative infrastructure provider (like an SNO or a power utility). Tier 3 ISPs offering VoIP services would either have a VoIP peering exchange or might have bilateral peering arrangements between themselves or might pay the incumbent telco as part of a bilateral VoIP peering agreement.

The creation of VoIP networks and their interconnection parallels the rise of the public Internet itself. Once common interconnection points are defined, VoIP networks will find their way there and peer with each other. It becomes cheaper to peer with those offering flat rate transit than those charging by the minute.

Once there is an IP-based numbering system (like ENUM or one of its equivalents), all VoIP service providers’ customers will be able to be reached by anybody else on another network. Let us say that you want to call a subscriber of (the imaginary) RealFastVoIPTelco in South Africa, you will simply dial +00 27 followed by the number. And at this point, the call can be connected to either a fixed or mobile network.

So what is enterprise peering? Simply two organisations connecting directly to each other’s network and agreeing a peering basis for doing so. Leaving aside the obvious advantages for corporates operating VPNs, in the developed world markets, two sectors (of some relevance to Africa) have been highlighted as likely adopters of IP PABXs: large government entities and educational institutions. In the developed world, these users often have high-end data equipment with bandwidth-intensive networks. Both also have multiple sites that need connecting. For example, ZDNet Australia reported last year that Deakin University had deployed 4,500 VoIP handsets.

The sceptics will rightly point out that Government in Africa is a slow-moving and ill-organised beast by comparison with even the worst of its developed world counterparts. But this is not a universal truth and the example of the “early-adopters” would put pressure on slower colleagues. Take the example of the municipality of Knysna in South Africa. Working with a private sector partner, it has created its own network that allows offices to talk to each other and council service users to ring in for free when making queries. Opting out of the incumbent’s relative expensive domestic calling will undoubtedly save money but it does not take much to imagine the same municipality peering with other networks at a local level.

All of the above is complicated and the speed of transition will be unpredictable within the developed world. However it is worth looking at how such a transition will begin to develop in Africa where the drag of historical obstacles is much greater.

In the first instance VoIP has to be legalised in such a way that a range of VoIP service providers are able to operate beyond the shadow of the incumbent. Thus far only Kenya and South Africa have really begun to offer this possibility but within the next 12-18 months there will be considerably more countries lining up to join them.

Then if there are plentiful users of VoIP, they need to see advantage in being able to peer traffic between themselves. So we need to identify the likely groups of organisations that will need to consider VoIP peering. These might include:

• Independent providers offering a voice service using broadband: In South Africa, this might include companies like wireless broadband providers, iBurst and Sentech. In addition, it could include the small number of Second National Operators (SNOs) who might also move into this area of business. The tiny number of carriers’ carriers would be well-placed to set VoIP peering exchanges as in the case of a company like Kenya Data Networks it will possess both an extensive fibre network and the corporate customer base.

• Traditional ISPs – both in the consumer and corporate market. Larger ISPs with a corporate customer base might see this as a service to their clients or find the faster moving asking them to do it. For the consumer ISP space, VoIP peering offers the opportunity to spread the customer base for “peer-to-peer” free calling which might not otherwise take off. However, it is worth noting that VoIP peering is different from existing IXPs in function but it may well be the same group of commercial and non-commercial forces that will bring about a VoIP peering exchange.

• Africa’s more far-sighted but increasingly hard-pressed incumbents. The competitive pressure of legalised VoIP will accelerate the installation of VoIP platforms and the efforts of carriers to find external international VoIP peering arrangements with a much wider range of carriers than has previously been the case. The number of NGN announcements will accelerate beyond a single hand of fingers in the next 12 months. The attraction of peering on a bilateral or multilateral basis is that it does not need cash, only traffic of a comparable level. This will avoid the kind of periodic “cut-offs” that some carriers have experienced, the most spectacular public one being Telkom South Africa’s closure of its fibre link to Zimbabwe’s incumbent TelOne for non-payment. Cynics who know the African market will say that these external cash arrangements on minutes are a fertile ground for a variety of sweeteners but market pressure is more likely to prevail in the long-run.

• Enterprises, universities and governments: Those corporates that have a regular traffic of calls with a clear and identifiable group of say suppliers will find enterprise peering a “no-brainer” as the IP PABX replacement cycle opens up more and more opportunities. Out of all of the things funded by Government, African universities have the most opportunities and strongest arguments for peering. Who could argue against a Malaria Research Institute peering with a Medical Training College? And to extend it, who could argue against a rural health clinic network VoIP peering with both of them?

The other side of VoIP peering is ENUM or some equivalent. Thus far Africa is listed in ENUM as a “public class continent” but has played no part in the initial tests. Thus far there is clearly little “push” in the argument for getting ahead with a test.

Pessimists familiar with Africa are always wont to stress how long it takes to get things done. But both the development of the Internet and telecoms in the continent runs against this trend but only at the expense of those running independent businesses doubling up as lobbyists. So it could be possible to see what is described above being something real on the continent within 3-5 years but we’re not taking bets on it. But the future’s out there and it’s waiting to be grasped.

The above article is a summary of the Introduction to Setting Interconnect Prices in Africa. For a full contents list and to order this report go to: http://www.balancingact-africa.com/interconnect.html