By Steve Jenkin

May 16,  2013


The Coalition has been very silent on one of its biggest and most invasive change to the NBN:

Every DSL-NBN subscriber is going to be saddled with three unavoidable out-of-pocket expenses. All in-house cabling changes have to be done by a registered cabler, it’s NOT DIY. You may go on e-Bay and buy a $50 VDSL modem, but it won’t work.

  1. Install a VDSL Central Splitter and new VDSL modem to get advertised speed. [$250-$500]
  2. When, not if, the copper phone service is turned off or you need a second line, install an NTD (Network Termination Device) as supplied “for free” for FTTP uses. The NTD contains two voice service connectors (via an internal “ATA”) and four data service connectors. [$500-$1,500]
  3. Either because you want a better service or when the DSL network is turned off in, say, 2022, you’ll pay to have Fibre run to your home. Charges by BT in the UK, whom it seems Turnbull is modelling his network upon, start at $1,250 and max-out at $10,000. That’s not a cap, they just won’t go further.

Step 1. Get VDSL working

The Coalition are saving a small amount, 5{17ac88c265afb328fa89088ab635a2a63864fdefdd7caa0964376053e8ea14b3}-10{17ac88c265afb328fa89088ab635a2a63864fdefdd7caa0964376053e8ea14b3}, in NBN construction costs of the NBN by forcing unavoidable costs onto householders. And not once, but three times, will ordinary householders have forced and avoidable out-of-pocket expenses. What’s not to like?

When a node is installed on your line, it will be “without disruption”. Your phone will still work, and given that your ISP can and will transfer your ADSL service from their exchange DSLAM to the node, your ADSL service will keep working.

Those nodes don’t just have ~150 DSL ports, they must also have ~200 phone ports, each with a filter that works for the two different ADSL and VDSL frequencies and a second internal distribution frame or patch panel. This makes them much larger, more expensive, and more complex than they need to be. More like the double-wide refrigerators that are Telstra RIM’s (with “Top Hat” conversions) than the very modest small-beige boxes of TransACT/iiNet in Canberra.

Not only are they forcing additional costs onto subscribers, they are increasing the cost & complexity (read “less reliable”) of the nodes. They could just follow the lead of the GPON FTTP rollout and provide an NTD and massively simplify the nodes and supporting network, but in the world of upside-down economics, a lower total cost doesn’t interest the Coalition.

If you want the advertised speed, you have to purchase a VDSL2 modem. Mr Turnbull has claimed they are “$50”, but not any brand-name models and nothing in the retail shops where you can get a least a modicum of pre-sales support and post-sales service and warranty. Perhaps if you bought 1,000 directly from overseas, they’d cost you $50. Looking at British sites, I see them in the $150-$250 range.

But it won’t work when you plug it in!

ADSL and VDSL operate with different frequencies and require different “splitters”. To get reliable service – nothing special, but just what’s been advertised and you’ve paid for – your best option is to install a “Central Splitter”. If you attempt to run “in-line” splitters, they are very likely to go awry and you’ll lose your service.

The Telco phone line into your house is cut and the “splitter” is connected before the first phone point. The splitter has two output connections: one for the DSL modem, the other for the phones in your house.

The cable up to and including the first phone point is the responsibility of the Telco. Only registered cablers are allowed to touch it, with some pretty draconian legislation applying to “line tampering”.

You’re not going to get a cabler out and crawling around under your house for an hour for under $100-$250. Buying a $50 VDSL modem now looks pretty irrelevant.

The majority of homeowners will take one of two paths to get VDSL working, for $250-$500:

  • sign-up with a specialist company (e.g. via large retailers: Harvey Normans, Domanye, Good Guys) that will supply and fit both a modem/router/wifi-base/firewall and the central splitter, or;
  • source the modem/router/firewall themselves and employ a registered cabler to employ the central splitter.

Step 2. OMG! The old phone is going to be disconnected, do something!

Another detail the Coalition is entirely silent on is the fate of the Phone Network. Currently there are many phone networks in Australia, not just Telstra, because of the ULL (Unbundled Local Loop) service retail phone providers can purchase from Telstra. They get to install their own telephony equipment in the Telstra exchanges and your line gets connected to that, not the Telstra exchange. I would expect all this Telco equipment is now “Voice over Internet”, not classical exchanges.

The problem is converting your analogue phone service to digital. This is done in a device called an “ATA” – Analogue Telephone Adapter. There has to be one somewhere: in the exchange, in the node or in your house.

What the NBN offers is not the sort of VoIP service that you currently buy from your ISP or a dedicated VoIP provider like Engin.

The NBN Co service is designed to meet Telco standards of clarity, reliability, availability and jitter. It does this by making phone traffic “high priority” (Class 1). Ordinary data traffic, (Class 4), is delivered on a “Best Efforts” basis.  This two-tier network design, and separate voice/data VLANs is well-known and tested: it’s how the market leader, CISCO, has delivered its VoIP phone for a decade.

There’s a second network detail: Voice packets are very small, very numerous and you’d rather not lose any. Normal data traffic, webpages, images and video, are carried over TCP in large packets, at a much lower rate. TCP checks for and resends lost packets. It also adjusts the rate it sends packets for the available bandwidth, and mostly doesn’t care about latency. They are different beasts, almost as different as is possible. Mixing them on the one network can only lead to heartache.

There are some other reasons you might want to upgrade to a multi-port NTD, as is supplied free to every other NBN network subscriber:

  • Multiple RSP connections
    • TV multicast or Cable TV providers
    • Education, health, human services provided access.
    • Controlled Work Networks for teleworking.
    • Multiple ISP’s with different access packages.
  • Direct or ‘native’ IP over ethernet.
    • High performance networking, not subject to overloading of PPPoE processor.
    • Lower cost voice services.

Step 3. So, you want Fibre like those lucky 3 million householders got in the first round?

At some point, the Copper Customer Access Network will reach the end of its economic life.

This will come because of two economic forces pushing in the same direction:

  • Revenue per User, ARPU, will be lower on DSL/FTTN because it offers a lower grade service and isn’t upgradeable. The service is being sold by the Coalition as “cheaper”, which will cause the consumer perception that it is an inferior service.
    • Consumers will not pay a premium for what they consider an inferior service. This is the basis for the Coalition pitch and backed up by all the technical and economic discussion.
    • The DSL/FTTN service must always be priced more cheaply than the comparable GPON/FTTP service, restricting its revenue-generating potential.
    • With Fibre services, it’s simple to create product differentiation based on speed. The Coalition seems to be offering on DSL/FTTN a single-product, “Whatever you get, you get”, without different prices for different speeds. The implication of publishing a rate-card with higher speeds is that they are obliged to supply them, not possible with the DSL/FTTN.
  • Network maintenance costs aren’t related to the number of subscribers using the network (“Fixed” not “Variable”), but must be paid for by the network users.
    • When there is a single Customer Access Network, there is no issue. All users share maintenance costs.
    • If customers are being lost to other Access Networks, the cost of maintaining nodes and copper doesn’t change, but the number of subs to share the cost decreases: as users leave the DSL/FTTN network, the cost per user of maintenance increases.
    • The copper network and remotely located nodes are the most expensive possible options for maintenance. The GPON/FTTP network will be 2-3 times cheaper to maintain.
      • While the DSL/FTTN network has over 66-75{17ac88c265afb328fa89088ab635a2a63864fdefdd7caa0964376053e8ea14b3} of subscribers, per-user maintenance costs will be lower than fibre.
      • Because of the high differential in costs, as soon as 30{17ac88c265afb328fa89088ab635a2a63864fdefdd7caa0964376053e8ea14b3} of subscribers have moved to fibre, per-user maintenance costs on DSL/FTTN will be higher and cannot ever come back.
    • This combination of much higher operational costs on an inferior service creates a feedback look, a vicious circle:
      • Costs go up, prices are raised, users leave forcing costs up more, forcing more users to leave…
The combination of effects, capped and limited revenue for an inferior service and very high operation expenses create the conditions for a “tipping point”. When the inferior DSL/FTTN service is more expensive to run and profits collapse, either the service provider will withdraw it or increase prices.
Even if there is a significant cost for on-demand deployment of fibre, this may not stop subscribers if their are businesses providing a “share the fibre” service for 30-40{17ac88c265afb328fa89088ab635a2a63864fdefdd7caa0964376053e8ea14b3} of the sticker price. Every distorted market attracts niche players who make a profit from differential pricing: technically, it’s arbitrage.
Remember the fall of the Berlin Wall and the subsequent collapse of the USSR and Communist Eastern Bloc: it was frighteningly fast. What looked like a solid, impregnable system hit a “tipping point” and collapsed within weeks and months. When the end comes for DSL/FTTN, it will necessarily be swift. Like the collapse of IBM’s business in 1991/2.

NBN Co, if forced to build a DSL/FTTN, would be acutely aware of the tipping point and the relative maintenance costs and per-network profit levels. As informed and skilful business players, they will pull the plug before they lose money. This is a guaranteed end-game, fast and total collapse, no matter what “vectoring” provides or what promises are made. The commercial question is: “How long till collapse?”.

Is it a wasted investment or will it survive long enough to return a profit for its $16 billion price tag and $800MM/year to run?? That’s over $1500 profit that NBN Co has to make for each of the 9M lines over a short and uncertain life. If you were doing the corporate plan, would you choose a life of 5 years or 10 years?

But only 60-70{17ac88c265afb328fa89088ab635a2a63864fdefdd7caa0964376053e8ea14b3} of the lines supplied will have paying services on them. The commercial situation is far worse: the per-subscriber, not per-permise, break-even is ~$2,500 profit per subscriber, or $500/year, well in excess of the ~$30/mth revenue expected. Even with a 10-year life, a $20/mth profit is required for the cost to be repaid. That’s not a bet I’d make…

Telstra, as the owners of the copper, will be paid yearly rental on the DSL/FTTN copper, which will cease if the service is abandoned. They will be acutely aware of this time limit and won’t put their already guaranteed payments at risk. Telstra will very carefully model the tipping-point and break-even conditions of a DSL/FTTN network and make a recommendation to its shareholders.

The Coalition cannot force Telstra to accept its conditions since it is no longer the majority shareholder.


The Coalition plan to move costs from the taxpayer (Government) to the taxpayer (subscriber) is only fiddling the books to achieve a political end. The taxpayer still picks up the tab, just in a different, much more expensive, way.

If you add the total costs, Government and subscribers, then their approach is much more expensive and anything BUT “cost effective”.
NBN Co in a mass-deployment would be extremely efficient, reducing costs 5-10 fold compared to “one-off” installs.  A large project reduces the overheads and delays to an absolute minimum and assembles complete teams of experts in the one spot to execute tasks at peak efficiency.
At the very least, this efficiency advantage, coupled with simpler, cheaper nodes without ATAs, would mean installing DSL-NTDs at rollout time would be cheaper.
Forcing each and every subscriber to individually arrange for their in-house cabling to be upgraded is the least efficient, most expensive option possible.