Monday, May 16th, 2011 by Simon Hackett
Deploying a ‘superfast’ national tail circuit network (i.e. the National Broadband Network) will highlight other bottlenecks in information flow that currently exist between voice and data service providers in Australia, unless those bottlenecks are also addressed. This paper highlights the need for appropriately symetrical peering arrangements to be enforced, for both voice and data services, between all NBN Service Provider networks.
The National Broadband Network (NBN) represents a huge intervention in the telecommunications realm in Australia. It is intended to replace the dominance of Telstra with a new ‘good’ monopoly that should be limited to operating in the wholesale realm alone.
While such a tail circuit network will address a critical part of the future communications needs of Australians, there are other bottlenecks that exist between consumers and the information sources they wish to access. Raising the tail circuit speed may simply highlight those other bottlenecks to information flow unless they are also addressed.
I will note two specific bottleneck points that seem to have escaped consideration thus far, and explain why some attention to these details within NBN policy would be of benefit to consumers.
Peering between NBN Service Provider networks
A consistent issue in Australian telecommunications has been around the notion of ‘peering’. In the telecommunications context, this refers to a scenario where two or more industry members support the smooth and efficient passage of information between their respective networks without adding artificial cost constraints; in other words, without one party imposing an asymmetric cost upon the other party in order for information to be exchanged between those parties.
The generally accepted charging model for such intertactions is ‘sender keeps all’ (SKA) peering – in other words, each party to such a peering connection pays its own costs to the point of interconnect and retains all income related to the flow of information across the link(s) concerned.
There are two key ‘peering gaps’ in Australia today, with one related to voice and the other related to Internet data.
Voice Network Peering
In the voice realm, Telstra (as the dominant fixed line voice carrier) has long imposed onerous, outdated, and expensive constraints (both technical and financial) upon other carriers wishing to achieve financially and technically efficient bilateral ‘local number portability’ (LNP) and the bilateral exchange of phone calls with the Telstra voice network.
The importance of this ‘voice peering’ in the context of the NBN is that the NBN will physically remove the existing distance based costs related to originating and terminating voice calls, and will replace that with the carriage of all calls to NBN customers via Voice over IP. Whether the customer end is implemented through the ATA (analog voice) port on the customers optical termination unit, or the call is delivered via the customers Ethernet port to VoIP equipment in their home, the marginal cost to deliver the call from the NBN Point of Interconnect to the customer falls to zero in this environment.
As such, it is logical to expect the replacement of the existing CCS7, distance based, circuit switched PSTN mechanisms (call origination, call termination, and local number portability) with the simple peering of each service provider voice network using Voice over IP.
This makes the use of SKA peering an obvious and appropriate mechanism in the voice realm for NBN based providers.
To date, industry requests to engage in this form of VoIP based, SKA voice peering and to implement LNP over the same framework with Telstra (as the dominant voice platform provider today) have been rejected, and it is therefore clear that some form of policy requirement for such voice interaction is necessary when the NBN is deployed.
Internet Data Network Peering
In the data realm, Telstra and others have historically formed the ‘Gang of Four’, a subset of Internet Service Providers who have refused to offer SKA IP network peering on a neutral basis and instead have imposed asymmetric costs on the connection and exchange of IP data between Telstra and parties not within the ‘gang’.
In particular, while other IP backbone providers in Australia peer extensively (via both bilateral and multilateral peering points around the county), the members of the “Gang of Four” (and most critically, Telstra and Optus) require other providers to purchase connectivity to and from their networks at commercial (transit) IP carriage rates.
These connections are expensive, and as such they are frequently not provisioned with sufficient ‘burst’ capacity to accommodate transient high demand for data flow. By contrast, SKA based peering links are generally provisioned with high capacity (and proactively upgraded) as there is no reason to do otherwise.
These peering links encourage efficient transfer of data between IP backbones, and ensure there is no artificial bottleneck created in the flow of Internet data between a customer on one network within the Gang of Four and a server attached to another network not inside the Gang of Four (or vice versa).
It is important to note that for some time now, the paid transit links between the Internode national IP backbone and the Telstra and Optus networks is such that we deliver more data to those backbones than we import. In effect we are paying to deliver data to those networks that is then on-sold to its customers.
Historically, the ACCC recognised this problem in 1998 and took enforcement action to require peering between IP networks. However, the determination made at the time was fatally flawed. The ACCC decided (incorrectly) at the time that the (then) dominant four players had to peer – but no requirement was placed upon those four to peer with anyone else, ever. All subsequent requests and attempts to peer with Telstra and Optus since then have been rejected by those network providers.
In 2011, only two of those players (Telstra and Optus) remain in a market dominant position, and both steadfastly refuse to engage in sender-keeps-all neutral IP network peering, despite the entire rest of the industry having engaged in extensive bilateral and multilateral peering arrangements for many years. This situation leads to higher costs and reduced network efficiency for consumers.
Why this matters
In the ultra high speed NBN environment it will become vital that both issues are resolved and that peering of voice and data occurs efficiently between all providers offering customer services across the National Broadband Network.
At 100 megabit customer port rates, a lack of sufficient peering capacity (and a lack of economically rational paths through which to expand that capacity in accordance with technical requirements) will rapidly impede commerce and consumer utility for high bandwidth applications.
Similar issues will exist with respect to fixed line voice services unless action is taken to ensure that voice networks are peered and that efficient number portability is made available to consumers ‘on demand’ across the NBN.
Indeed the mobile number portability regime in Australia only exists today due to such an intervention to force it to happen. A similar intervention is clearly necessary in the NBN voice services realm. It will not happen by itself.
Addressing these policy gaps
Having indicated the existence of these problems, the question becomes one of how to ensure that efficient neutral peering in both voice and data do occur between all retail service providers that operate on the NBN. Where could a policy be defined and made to operate, in order to achieve this outcome?
A number of points exist at which such a policy could be defined and required of NBN service providers. These include:
However it is achieved, it is clear that technically and financially neutral peering of NBN service provider networks (voice and data, including local number portability) are essential parts of ensuring that the NBN delivers the full benefits to Australians that we all expect of it.
This paper was first published as ‘Peering Policy Gaps with the National Broadband Network’, in the Telecommunications Journal of Australia, May 2011, Vol 61 N1. This entire May 2011 issue of TJA is available to online TJA subscribers here : http://www.tja.org.au/index.php/tja/issue/view/14