Is the Turnbull NBN A National Disaster?

There are so many things to care about in the upcoming election. Issues of significant moral conscience, national stability, and economic futures. Issues that are set to shape the future of Australia and that need to be seriously considered and managed.

But one issue that keeps popping back up for discussion is the National Broadband Network. In particular, many people (me included) are criticising the current federal governments MTM (Multi-Technology Mix) approach. This is an area which I think I actually can speak with some authority. In a professional context I am a customer of an NBN aggregator, and also run a Carrier business that competes in the NBN rollout space. I’ve worked in the telecommunications industry for 16 years, and telecommunications is a significant part of what Real World does.

But to do this I need to lay out some information that is important to understand in the context of the discussion.

The NBN was already broken

(because it’s too expensive to get access to and deliver services over)

The NBN was designed to create infrastructure that would benefit the whole of Australia, to foster growth, economic stability and a technological future. It relied on cost modelling that subsidises regional Australia with comparatively high connectivity costs by building lower cost services in metropolitan areas. The project was always going to be difficult to be cost neutral – but it was infrastructure which was to have a 20 – 30 year life, and so many argued that the investment was worth it for the future.

Like any product that is sold, it needs to have a cost model and fee structure. When you work out the price of an NBN retail service (i.e. what you buy) there are three principal components to the cost.

  1. The NBN Access Charge. This is the cost for an NBN “port” in your house, and the path between your house and the NBN Point of Interconnect.
  2. The NBN CVC (Connectivity Virtual Circuit) charge. This is the cost for your ISP to buy access to a “port” in a Point of Interconnect and is purchased in “Megabits per second.”
  3. The “Backhaul” charge. This is the cost of your ISP taking the service from the Point of Interconnect back to their network.

So how much does that actually cost, and what’s the issue?

  1. NBN Access Charges vary, but a 25/5Mbps costs $27.50 per month.
  2. NBN CVC Costs $17.50 per Megabit per second. So, for a provider to provide 25 Megabits per Second of speed, they need to buy $437.50 worth of access. This access can be shared between multiple customers, but this creates “contention” – just like when you have two people having a shower at home using the same hot water system. Most networks contend their access at 50:1 – so for every 1 Mbps they buy, they have 50 customers trying to use the same space.
  3. Backhaul charges vary depending on the location of the POI, but range from 30c per Megabit per second up to $8 per Megabit per second.

Now these numbers don’t look like they are “too” big a problem until you start to work it out. Telco’s used to buy ADSL access from Telstra per State or Territory around Australia at $35 per Megabit per second (the price is cheaper now). There is now 121 places that ISPs can buy NBN services from. ADSL services on average run at about 1.3Megabits per second.* Because NBN services are faster, the average usage jumps on them from 1.3Megabits per second to 6 Megabits per second. This means that the average cost is 3 times that of an ADSL service – but the retail price has stayed the same. In addition, there is extra costs associated with delivering the services as there is more places to interconnect, with significantly more variation in backhaul costs.

Fundamentally, this makes the NBN “unaffordable” to sell, without offsetting your internet service revenue against higher cost “business” services or subsidising NBN with other technologies such as Voice over IP, Video on Demand or Mobile.

Take home message: The CVC cost and number of POIs are too high and too many to make the network affordable to use.

Copper isn’t a big deal today

Memes like this annoy me a bit.

Copper NBN is not “slow”. It’s not as “fast” as Fibre, but it is still fast. The technology (Fibre to the Node) is good science and has a definite place in telecommunications. But that’s not the problem with it. Most individuals and businesses don’t need the speed that Fibre offers today. In most cases, 25Megabits per second (a reasonable average VDSL speed) is perfectly reasonable. There are obviously some exceptions to this rule. The problem is not that VDSL is bad today – the problem is that VDSL is not a long term solution.

Why? Because the copper network is not that great. Copper is a metal, and it corrodes and degrades. It has a limited shelf-life (20 – 30 years from what I’m told). Physics shows us a lot about how electrical pulses can be sent over copper, how they interfere with each other and at what rate the degrade. And in many parts of our country we are already seeing the effects of the age of the network and degradation. In addition, copper services can’t be shared – so one “pair” of copper cables equals one service.

In comparison, fibre is glass or plastic. It has a shelf-life of 60 – 80 years. Light travels along it just below the speed of light. The major issues with fibre are that it gets cut or dirt gets into the connectors, which can be cleaned with an air gun. Fibre services can be shared as multiple colours of light can be sent down the same strand of fibre allowing you to run multiple seperate services on one piece of glass. The downside is that you have to run new fibre cable to each premises you want to service.

Is copper cheaper? Well, yes it is. Because we already have a lot of it in the ground, and so we can install a “node” and connect the copper that is already there to the node. But at some point, the cost of installing, running and maintaining nodes, upgrading or repairing the copper outweighs the cost of installing fibre. And we are still needing to run fibre to each node; so there is still a lot of cost involved in getting the fibre there.

Take home message: FTTN (or the Copper NBN) is not a “slow” solution today. But it is going to cost a lot to keep running and won’t scale into the future.

The cost justification is wrong

The Senate Estimates Committee has made it very clear that infrastructure spending for the NBN is a 4 year election-based decision. Unfortunately, any telecommunications network is a 10 year investment at minimum. When you make decisions about spending on an election 4 year cycle, it makes sense to choose the option that will best benefit your bottom line over a 4 year period. But the down side is that you push the cost of running and maintaining that network 4 years down the track – and ultimately onto our children, or children’s children.

For something that is of such national importance, it makes sense to consider the long-term economic costings, which are heavily geared towards Fibre being cheaper in the long term because the density and maintenance costs are lower.

Most of the numbers being released by politicians suggest that the MTM NBN may save us much as $30 billion, although this number changes regularly because no one really knows how much either network will actually cost to build. The industry estimates seem to suggest the numbers are closer to $10 – 15 billion – which is about 20% of  the overall cost of the project. [Note: this paragraph previously stated grossly inaccurate cost savings.]

Comparatively, Telstra’s agreement to maintain the copper network  it sold to NBN is worth about $80 million per year; on top of it’s Fibre maintenance costs. (This cost also includes new connections, so arguably the maintenance costs are only a portion of this final figure).

My concern is not about the incidental cost now – my concern is how much are we going to have to pay to keep the network running 5, 10 or 15 years from now. Are our Children going to be having this conversation all over again, being forced to spend the same money all over again to keep the telecommunications infrastructure up and running in this country to deliver the services they need.

It just doesn’t make sense to save a few dollars now for the sake of a much larger cost later.

Take home message: We are wasting money by spending so much on a Copper NBN when we are going to have to maintain and replace it in the future. We should just do it right once, knowing the decision we’ve made will last another 60 – 80 years.

Will the NBN influence my vote on Saturday?

Probably not. I’d like it to, but there are so many other things that are big issues for our nation such as:

  • the treatment of Asylum Seekers
  • the need for greater Domestic Violence Prevention
  • the state, health and protection of our Environment
  • the role of Gender Identity in our education system
  • the definition of marriage
  • the funding of Tertiary education
  • the state of our Nation’s economy
  • the impact of Brexit on Australia
  • the need to protect free speech across our country
  • the global impact of terrorism
  • the need to foster innovation and development to grow our economy
  • the need to foster and grow small business
  • the continued provision of quality health care

… and so many more big issues that weigh on my mind. But let’s at least go into this with our eyes open, understanding the issues and be prepared to meaningfully discuss what is going on.

The NBN may in fact be the burning issue that gets your vote – but just remember that it is one of many big things that are impacting our nation at the moment.

* This isn’t your peak speed, but when you plan your network, you can guess that your customers will use about that much bandwidth and build your network from there.

For some further reading around the cost model this article is helpful on cost modelling. This presentation from AusNOG last year also has some good analysis of the speeds required for NBN services and the impact on CVC.

Thanks to Karl Auer for helpful criticism and comment which has resulted in a few modifications and clarifications to this post. Karl has also pointed out that a number of the elections issues I’ve identified are helped by an excellent, fast, national broadband network, that is ideally delivered over Fibre. I think there is a lot of truth to this statement. Thanks!


4 thoughts on “Is the Turnbull NBN A National Disaster?

  1. Andrew,
    A couple of thoughts to add:
    – Many of the early Optical Fibre cables installed in Aus. (~30 years old) are now so brittle that they cannot be spliced and will soon need replacing. Newer cables claim to have a greater life span, but that remains to be seen in the real world. This means that if we replace all the copper cables with fibre in a short period, at some time in the future, they will potentially need replacing in a short period. It also means that replacing copper cables that are in good condition, before increased performance is needed, unnecessarily brings forward both the initial install cost, and the replacement.
    – The nodes that are installed for the FTTN, can be used as nodes for FTTP, usually by just switching the cards on the customer side. This also means that individual customers can be upgraded if required.
    – in reality every technology we use to connect is fibre to a node, which is shared backhaul, then an access technology. Some access technologies like GPON, xDSL, have unshared channels from the node to the customer, others like HFC, WiFi, LTE, WCDMA have shared bandwidth from the node to the customer. Which one makes sense for depends on a number of factors.
    – Just about every hospital, school, and medium to large business in Aus is already connected by fibre, often at speeds much greater than the NBN. 10Gb/s is not uncommon, and 100Gb/s is available. Many small businesses also have fibre, where they have judged the installation cost worth while.

    Fibre is obviously the best available technology for connecting, and for new installations is now a comparable cost. Where the copper is able to support 25Mb/s and higher, I’m not convinced that the taxpayer should be footing the bill for upgrading yet.


    Liked by 1 person

    1. Definitely some good points there.

      You’re definitely right that the fibre currently used hasn’t had enough age to determine if it’s going to have the same problems 20 – 30 years from now. Certainly the fibre manufactures and scientists who have been working on this are of the opinion that we’ve adequately pushed out the life of the fibre; but time will see.

      The nodes that NBN are deploying are the previous generation of ALU nodes (the current generation doesn’t support the VDSL line cards) so are limited to GPON, rather than NGPON or 10GPON. This probably isn’t a big limitation in the short term, but it will mean that nodes will, at some stage, need more than a line card swap because of the 2.54Gbps downstream bandwidth per PON port limitation. At 1:16 or 1:32 this isn’t a big deal provided you are only offering 100Mbps services at TC4, but once you start to play with TC2 or TC1 services, or need more bandwidth then I think that changes the provisioning dynamics.

      You’re definitely right that many businesses are “shortcutting” the NBN rollout and putting in fibre themselves. The cost for direct fibre varies from $500/month up to several thousand dollars per month, and installation costs vary from $0 (in metro) up to several hundreds of thousands of dollars, depending on the quality of the pit infrastructure and underlying duct network. There are also some significant differences in how this is generally deployed, with the fibre being installed into businesses being active fibre on dedicated wavelengths, rather than CWDM in most cases. (AAPT’s Fibre 400 is a noticeable exception to this where they appear to be doing 1:8 splits on GPON optical). Because of the technology differences, I’m not sure that this service is directly comparable to ubiquitous fibre broadband Australia wide, but I know the opinion is varied. Also, for every 10 locations we try to connect fibre to, around 3 can not be serviced.

      I think by comparison, we also see a much higher fault rate on FTTN services over FTTP services and customers get significantly lower speeds than the “theoretical maximums” – i.e. customers “pay” for a 100Mbps but only sync at 60Mbps. This is quite common, and is the reason why NBN FTTN services are now sold as “up to 100Mbps”. The copper quality and cross talk are the main contributors for this. Is this a big deal? Not sure; but for 20% cost difference, why not just do it “right” the first time?


    2. Also, I think I’d be happy with an approach that actually assessed copper quality to determine if FTTN would be feasible in an area; if the copper is good, if there is a low fault history, and if the pillars are well maintained then it is probably a good low cost option. Such assessments don’t occur at the moment in the technology model.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s