One of the most important questions in the NBN debate is the location of the network’s points of interconnect. Even though the ACCC held only a relatively brief inquiry into NBN Co’s point of interconnect plans, the inquiry attracted nearly 30 responses.
So it’s an issue that the industry sees as important – but what’s it all about, and does it matter to the end user?
What is a POI?
A point of interconnect is a place where two networks pass traffic to each other.
Imagine that you’re an iiNet ADSL customer. The wires that connect you to the exchange belong to Telstra, as does the exchange: what iiNet owns is the DSL equipment within that exchange.
In this case, the POI defines the point at which traffic ceases to be handled by Telstra, and iiNet assumes responsibility for the traffic.
In the NBN, customer traffic will be collected “at your door” by NBN Co, and will be passed from NBN Co to your retail service provider at a point of interconnect.
Just how that interconnection takes place lies at the heart of the POI debate.
At one extreme, the NBN Co fibre serving areas could be considered discrete “bubbles” of connectivity. A retail service provider would buy access in Tamworth, but to connect those customers, it would build or buy a link from Tamworth to its own network (such as to a data centre in Sydney).
At the other extreme, NBN Co could create a nationally-integrated network, buying long-haul fibre services (because except in special cases, NBN Co will not build its own backhaul) to connect all of its FSAs together. Retailers would buy a single connection to NBN Co, and thereby gain access to a national network.
The two extremes – and all possible permutations in between – are defined by where NBN Co decides to put its POIs.
Retailers are already accustomed to dealing with a large number of POIs. There are roughly 500 in the Telstra network. Pretty much any exchange in which Telstra’s competitors have installed ADSL equipment is designated as a point of interconnect. In a large number of regional locations, that means buying a bit pipe from Telstra, because Telstra is the only carrier with a connection from that exchange back to the ISP’s own footprint. Some regional cities are served by a handful of fibre owners, giving the ISP more competitive choice; and major capital cities are well-contested by a large number of fibre owners.
NBN Co’s position
In the discussion paper it put forward in October, NBN Co stated its preference to have 14 POIs, distributed around Sydney, Brisbane, Melbourne, Adelaide and Perth.
The key argument is cost. Remember that NBN Co is building an entirely new network: each POI it has to build feeds into the build cost.
Fewer POIs means lower cost, and NBN Co would pass on the saving to its retail service providers.
But POI location doesn’t only affect NBN Co: there are at least two other communities with a direct interest in this part of the network architecture: service providers, and companies that already own fibre backhaul.
The ISP’s view of the POI
Whether an ISP wants local or aggregated POIs depends on its network footprint. If you’re a national service provider, or if you’re located in the capital cities where NBN Co proposes to built its POIs, then this highly aggregated network looks like a good idea. With just a handful of connections to the NBN, the national or metro retail service provider would gain access to all of the NBN’s fibre serving areas – whether they’re in cities or out in the regions.
On the other hand, if you’re one of the dozens of ISPs that’s built a customer base and reputation by focusing on the needs of various regional areas, the aggregated POI option doesn’t look so hot. If your data centre is in Tamworth and the POI is in Sydney, you’ll need a long-haul connection to get to the POI (unless, as would be likely, the Tamworth retailer decides to relocate its data centre).
The ISPs also have to look at their cost base. If you’re a national provider with a large number of POIs, the aggregation of traffic into a single POI reduces the cost of deploying your services.
And at the same time, larger national providers who already have data centres in capital cities would get a relatively easy way to give themselves a nationwide regional footprint: the connection they’ll need to the NBN POI will give them that footprint.
The fibre owner
There are only a few companies with truly national long-distance fibre networks – Telstra, Optus and Nextgen reach from the East Coast to Perth, and AAPT has fibre in the eastern states. A number of fibre owners also have long-haul links within Queensland, NSW, Victoria, South Australia and Western Australia.
These companies are vitally interested in the NBN’s POI architecture, because it affects the future value of their networks.
The concern of the long-haul fibre owner is that it might find itself with empty links.
Let’s imagine that Tamworth is served by fibre from Telstra, Optus and AAPT. If NBN Co builds 14 capital city POIs, then t will need fibre to get traffic from Tamworth to the POI. If NBN Co decide it needs redundancy, it will buy services from at most two of the fibre owners who have connection points in Tamworth; the third will be left “out in the cold”.
On the other hand, if the POI is in Tamworth, retailers offering services in that area will need to connect their traffic to the outside world – and the more retailers decide to link to the Tamworth fibre serving area, the opportunities the fibre owners have to sell links back to the city.
In other words, a highly aggregated POI model increases the risk to the fibre owner: there’s only one buyer in town, and therefore only one sales opportunity.
Do the NBN POIs matter to me?
The government has instructed NBN Co to provide uniform, national wholesale prices. The aim is to make sure that someone who wants 100 Mbps access in Wagga Wagga doesn’t have to pay a price premium based on location: the retailer will pay the same wholesale price for that connection as it would pay in Sydney.
The highly aggregated POI model simplifies the job of creating that national wholesale price, since NBN Co would have control of the greatest variable: the fibre price. So on that front, the aggregated POI model looks like a positive for the end user.
But even for end users, there’s still a trade-off involved. If the regional specialist is placed at a disadvantage, then so are its customers (unless they move to a larger-scale provider when the NBN arrives). If the POIs are disaggregated, then both NBN Co and the larger providers will have to carry a larger cost base, but there will be more room for local competition and innovation.