Friday, 26 April 2013

NBN: Why FTTN rental must rise in a dual FTTP/FTTN world

The NBN deal the Coalition is offering to the general public is:
We're going to build a network past 12.2 million premises, saving $250 out of $3,000. In exchange, it's going to cost all of the unlucky 9 million homes, $5,000 each to ever get full Fibre. 
And they make an implied guarantee about Affordability: we'll keep the price of DSL cheaper than Fibre, forever, which is really 15-20 years.
Only that's not nearly true.

If on your street of 50 houses, 14 of them have Fibre and 36, including you, are on DSL, then all the costs of maintaining the whole of each network must be shared between those using a network. This is much more than the last 350m that runs down your street.

Whether you have 1 or 50 people connected to a network in a street, the costs of running the whole network, to connect, manage, bill and maintain for those houses, stays the same. The variable costs of providing any network connection to a house is almost, but not quite, nil, its the small amount of electricity used. The fixed costs, those that don't change with the number of services connected, are more than 99% of the network costs, operational and capital.

This is the economics of all Telecommunications networks, including wireless, and what made two Cable TV networks serving the same addresses pure madness: the additional cost of another user is too small to measure or bill. Telcos must spread their costs across a large number of users, or go under, as Cable TV did here within 5 years of being built.

Those 14 people on Fibre get to share the whole costs of their network while the 36 on DSL get to share the whole costs of their network. Funnily enough, those costs are not all that different, 10-25% different at most.

In the beginning, the share of costs of the "lucky few" on Fibre is two and a half times more than their  DSL-connected neighbours. The company that owns the DSL network can charge less and make more profit! That's "a beautiful set of numbers".

But what happens in 10 years when the situation has reversed because 11 of the 50 houses have converted to Fibre?

The situation is reversed: the "lucky majority" now get to share the whole network costs and their costs are much, much less, making the operator now able to charge much less for Fibre.

What's happens for the DSL houses is their share of maintaining the network increases by 250%.
The operator has two choices: absorb the costs of pass them on...
Any rational business must pass on the costs, raising DSL prices considerably, for a less capable service.

Of the 14 of 50 houses now left using DSL, many will start to wonder if its a good deal anymore: slower and more expensive! The service will also be much less reliable for technical & commercial reasons. Operators will direct their maintenance dollar towards the more profitable service, Fibre by then. DSL performance and reliability will decline because of falling revenues and margins.

If just one more house "defects" to Fibre, that increases the costs to those left by 7%. Not that much, but 275% of the original share of costs and 285% of the share of the other group.

The next house to go, "and now there were 12",  adds 8% to the latest cost, or up 300% on the original cost and 315%  the share the other group pays.

For 10 houses, they're paying 360% more than when they started and 400% more than the other group.
For the last 5 houses,  they're paying 720% more than originally and 900% more than the others.

Rational customers will flee the old network as charges start sky-rocketing, compounding the problem.

The only questions for the network operator are:
  • When do we give those stuck on their old service a "special deal" to step-up to the new service?
  • When do we declare the old service "no longer economic" and shut it down?
This is a cascade change or "tipping point" problem. As the more desirable service becomes cheaper, the barrier to entry reduces so it becomes more desirable and cheaper etc etc.

At some point, the more desirable service is also the cheapest, then everyone jumps on-board, or is forced to jump when the operator kills it. A good operator will maximise their profits and abandon a failing service well before it starts making a loss. The decision point is when the Gross Margin (%) on the new service is more than the Gross Margin on the old service.

Normally a tipping point is 40-50% of a population. Fibre starts at 28%::72%, if only 12% of premises switch to Fibre, the ratio is now 40%::60%, well within the tipping-point range.

The Maths is simple and relentless, the business decision is unavoidable.

The only question for Telstra and NBN Co is:
How long before the DSL/Fibre tipping-point is reached?

The time to market collapse is very important to both businesses: it sets the maximum time they'll agree to rent the copper, which determines the yearly rental price.
Telstra has to get $3 billion in rent (NPV, PostTax) to make the new Coalition deal as good as its present deal, NBN Co needs to pay the least rent it can to make a profit.

Even at $5,000/premise to "Got Fibre?", one in eight houses can reasonably be expected to switch in the next 10 years for both its utility and for the effect on house value.

The accounting depreciation life of DSL nodes and lines may be 20 years, but it's extremely doubtful the service will get to run 10 years, especially if it turns out there really are services that only work over Fibre.

If I was a Telstra shareholder, would I accept any deal to rent the copper & not stick with the current deal?
Absolutely not, it is way, way too uncertain for me.

What if a Coalition tried to force their hand with regulation or legislation? It'd be challenged in the Hight Court. The additional time would scuttle their plans: they can't afford either losing or winning such a case.

The Coalition has to strike a deal with Telstra, and it has to be to be "rent not buy".
Telstra doesn't have to change, it already has a great deal.

I don't play poker, these types of plays are too much for me. How this will play out is outside my experience and knowledge.


  1. Steve, Just clarify the "2 seperate networks" point. If FTTN connects fibre into the current street, why isn't that just a single hybrid fibre+dsl network? Why is there 2 seperate systems of maintenance?

  2. Steve, Are you saying there are 2 seperate networks with 2 seperate maintenances and 2 seperate telcos? Don't quite understand that.

  3. NBN Co under the Coalition will build and operate *four* distinct digital customer networks, plus look after special services (fire & security alarms, traffic lights, ...) plus run a Phone network plus have something to do with keeping the last copper phone services going in remote/rural areas. As well they have transit networks, other monitoring/control, remote management, billing, provisioning and common functions. Backhaul from the PoI's is provided by other businesses.

    The 4 digital networks are:
    - Satellite
    - Fixed Wireless (mobile phone 3G/4G with fixed antennas)
    - GPON & active Fibre to the Premises, and
    - VDSL2 from Nodes, aka Fibre to the Node.

    The only point all 4 digital networks touch is at the Points of Interconnect, where every customer service looks identical: a Layer 2 'bitstream' with VLAN's. (802.1QinQ)

    The technology, equipment, network topologies and *fibre* used in the FTTP & FTTN networks are completely different. Despite FTTN needing 80,000km of Fibre, you'd install 12-72 single core, not the 846-core ribbon/tube FTTP uses.

    The only substantial asset than can be reused from a FTTN is exchange rack-space.

    How to upgrade FTTN to FTTP? Cheap-and-nasty off the node is really, really stupid and ultimately 5-10 times more costly.

    The only sane upgrade of FTTN to FTTP is an overbuild... What's the *right* number of Fibre Optic network architectures to support, repair, replace and maintain? Just One...

    - 1 company, NBN Co, builds, owns, operates, maintains 4 separate Customer networks. 2 of which are fixed-line.

    - Many Retail Telcos, only one Wholesale provider of Customer Access Network

    - modulo whatever craziness is allowed with HFC

    The maintenance teams for Copper, Nodes and GPON/Fibre have completely different skill sets, duties and areas of operation.

    They may at times work in the same ducts, pits and pipes.

  4. Just sticking to your 50 house example to further clarify this. (And not trying to create mischief here, I've read through ALL of your blogs posts and am in total agreement with the general thematic.)

    Assuming a Coalition FTTN deal, the 14 houses on fibre would have been on-demand connections at $5000/per, and presumably done, as you say: "Cheap-and-nasty off the node". That is, connected from the cabinet using seperate fibre line into a $400 NTU (this also at customer's expense).

    So as far as that goes, it's a terrible deal for the customer viz a viz Labor's FTTP freebie-deal.

    But this is where it gets a little confusing. You're saying that from the out-point at the node cabinet, we're dealing thenceforth with 2 seperate networks, fibre and vdsl. Requiring 2 seperate skill-sets to maintain (thus conjuring up the ludicrous image of 2 sets of guys wearing different coloured vests and tool-boxes working in a confined space getting in each other's way).

    I'm visualising your 2-network argument this way:

    network 1: a red line representing fibre going into the node cabinet and another red line representingn fibre going out from the cabinet into the customers NTU

    network 2: the same red line representing fibre going into the node cabinet and a blue line representing copper going into customers NTU or vdsl router

    (And, incidentally, what's not clear is whether we're talking about a universal-type NTU handling both scenarios, or an NTU for the full-fibre case and a normal router for the vdsl case.)

    I can see there are definite inefficiencies and cost-overlap from having the 2 systems co-existing in the example street. And that's probably enough right there to kibosh an FTTN system with on-demand FTTP upgrades. In fact I'm sure it is.

    But from my non-expert perspective, the red line of fibre going INTO the cabinets remains common to both scenarios, and therefore I'm slightly worried that you're EXAGERATING the costs of maintaining the node-ntu part of the connections, even though conceding that this part has 2 different sets of characteristics according to the 2 scenarios.

    Your argument only works 100% if indeed the entire red-in red-out fibre connection, and the red-in blue-out copper connection have to be judged as TOTALLY SEPERATE NETWORKS REQUIRING SEPERATE MAINTENANCE PROTOCOLS. In other words, are practically mutually exclusive. If that's the case then of course you have a very strong argument indeed.

    Anyway, congrats on the whole series of posts, which are far and away the best on the web on this subject.

  5. @dedalus Yes, the networks are totally separate, they part well upstream of the nodes.

    The nodes *could* handle a few fibre connections, but you'd direct connect with GPN if you got above 20% premises on fibre in a distribution area (200 premises).

    The Coalition NBN Plan continues the current (free) GPON FTTP build so 28% of premises are covered by 2019 or 2021. Figures & times are muddled.

    The 14 of 50 got there for free. I posit that 6 of 50 need to swap to fibre for a tipping point to be reached, and they will have to pay for the upgrade.

    Will the $400 already paid for an NTU on DSL be lost? Probably.

    The remaining 30 houses will also have to pay for their upgrades, though I'm thinking it will become cheaper if major conversions are done as the copper becomes uneconomic.

    How separate are the two networks? Are the nodes built to throw away in 5, 10 or 20 years?

    To get an upgrade to Fibre, a premise should be connected to the large-scale fibre ring, not the node, GPON not the FTTN.

    Each of the 121 PoI's, Points of Interconnect, connect 100,000 premises to the Retail Provider networks.

    Active Fibre links run out into the field: expensive high-cacpacity ethernet. Say 100Gbps to begin with. WDM (multi-colours) can be used for more capacity.

    The next level covering up to 3,200 premises, is called a Fibre Area Serving Module. This is where the fibres from individual services terminate. The FASM houses the *passive* Optical Network equipment (PON in GPON).

    The FASM connects to up to 16 FAN's connecting 200 premises in a "distribution area". Somewhere, maybe at the FAN are "optical splitters": many premises share a single transmitter. This reduces the amount of upstream fibre.

    Nodes will connect around 100 premises from what we're told, maybe as few as 25. Copper Nodes and Fibre DA's don't map well onto one another. The network "topology" - how each premise is connected is different and incompatible.

    Fibre is run in long (1-5km) continuous loops, or 'rings', copper is run in very short (400-800m) lengths in a star or hub-and-spoke.

    The 'magic' of PON is twofold:
    - by using simple optics ('passive') a single GPON transmitter can handle upto 32 premises. Around 10-15 premises at 100Mbps can share a single transmitter, I believe.

    - the only 'active' components (read 'electronics') are in the FASM and premises, nothing in the field. Changing to 10Gbps means upgrading the FASM transmitters, then as customers want higher speeds, replacing their NTD.

    The Coalition claim their ISAM nodes will connect copper and fibre, and that's not technically wrong.

    As a network operator needing to replace all the copper, you'd rip out the nodes as well as the copper because *active* networks done this way are many times more expensive to own and operate.

  6. @dedalus thanks for your kind words. I'm glad my rambling comments have been useful to someone.

  7. That's a much improved explanation Steve. Of course, I should have twigged to the fact that you were discussing a typical street in a previously rolled-out area where 14 users had already taken up NBN fibre.

    One little point is very significant as an anti-FTTN argument: and that is that end-user may have to pay TWICE for the NTD. Since I'm assuming from what you say that the NTD for vdsl is a different beast to that for fibre, it follows that if an end-user had signed on for the initial FTTN, for which he had to pay $400 for the NTD, and then LATER decides to ugrade to fibre for, say, $5k, the poor sap will have to fork out another $400 for ANOTHER NTD.

    Am I wrong or am I wrong? I think that point needs to be validated from source. It's a really powerful point.

    The other thing that needs clarifying is a phrase I'm sure I read somewhere in some Turnbull guff, which is the phrase "if economic". Now there is a doozy, and I reckon it means that NBNCo mark 2 will upgrade you to fibre "for several thousand dollars" (using Malcolm's exact words here from the Jon Faine interview - link on my website, provided - and here's the rub - that it's "economic" to do so.

    OK, I'm implying stuff here, but it sounds plausible. So, say you're the trusting fool (who bought T2 shares for $7) living at the far end of a very long street. So NBNCo is going to dig under 3 sets of traffic lights and the kindergarten playground to lay out fibre for little old you? (Exaggerating for effect.) More likely you'll be PLACED IN A WAITING LIST, and when the list becomes "economical", then you'll get your fibre ... maybe ... zzz .. don't count on it

  8. dedalus...

    Some links for you, diagrams might help people understand.

    Selected pages with diagrams from the 2010 plan
    the full 2010 plan

    The Mythical Street I described may come to exist. Up one end, the copper needs replacing, so instead they get full Fibre, down the other end, they get DSL via a Node: all at install time.

    No upgrade bought by those lucky enough to get Fibre. Meanwhile, those initially connected to the Node do get to pay for an upgrade.

    The other side of this was an-example-in-the-small of the whole network.

    NTD's?? We just don't know. Coalition has mentioned them, so we have to presume they aren't included.

  9. Steve, I follow your blogs with great interest. Although no longer involved, I used to teach network technology, as well as having worked professionally with and in communications, both military and civilian, and so they make great sense to me. Like 'dedalus', may I also congratulate you for your succinct, accurate and non-partisan information. It is a pity that your arguments about the idiocy of the proposed coalition proposal aren't given far greater exposure! They certainly are the most lucid available on the web.


Note: only a member of this blog may post a comment.