In my usual Saturday morning ritual, I donned my dressing gown, sat down at my desk with a coffee in hand, & read the morning’s news & tweets. A shooting at LAX, some fundamentalist Christian homophobic rant from a footballer (quelle surprise!), & more news on the ALP’s internal struggle with carbon pricing. Not exactly a news heavy day, easy enough to get across.
So I decided to do some digging on FTTN deployments worldwide, mainly because it’s been weeks since I received a single alert email to FTTN articles going up (apart from some oil company whose stock ticker is FTTN). Which is odd since every few hours I get an FTTP related alert, & every day or two one on the UK’s FTTC (it’s the only FTTC deployment I can find actually).
First I thought I’d look for FTTN deployments worldwide, none are active or even planned from my searching. Most of the results point to Turnbull’s policy/plan, the rest to documents by communications equipment manufacturers. Down the rabbit hole we go!
The first document I happened upon is a rather interesting report from 2009 by ADC Krone, now TE Connectivity, titled The Economics of FTTN vs FTTP. Keep in mind, this is a 2009 document, & we’ve seen a massive drop (60%-90%) in the costs associated with FTTP deployments since 2008.
Krone emphatically state:
At the end of the day, the key consideration in choosing which architecture will work best for any provider boils down to bandwidth – now and in the future.
A rather pragmatic statement from a company in the business of making money off service providers, then again, Krone is one of the few telecoms companies who haven’t pumped billions into DSL research. Krone make connectors for both electronic & light communications (copper & fibre), so whichever you choose, they will still make money.
An interesting point made in the report is that for any copper based solution to be successful, bonding may need to be leveraged, causing a mess of new headaches. Bonding is a rather simple concept, instead of using a single pair (2 wires) for your service, multiple pairs are used, generally 2 or 4 pairs (4 or 8 wires). This also means multiple ports are used on the DSLAM, as the report states:
There are other issues that must be considered to successfully implement this technique. For example, this technique requires using more of the DSLAM to serve fewer customers. With that in mind, a 192-circuit DSLAM may only be capable of handling 96 customers if multiple HDTV services are demanded at each residence. Also, if the distribution area requires resectionalization as reachable distances decrease (for example, deploying VDSL will require new cabinets for reaching customers outside a 3000-foot perimeter), the deployment of additional adjunct DSLAMS may be necessary. Furthermore, if existing cross boxes do not have the necessary binding posts to support them, this initiative can quickly become very complicated and expensive.
The Krone article’s conclusion makes a point that I feel Turnbull is continually downplaying:
However, there are still many unanswered questions about bandwidth demand over the long term that cannot be ignored. FTTN architectures will continue to have bandwidth limits that may be exceeded – perhaps in five years or less. On the other hand, FTTP is certainly a more future-proof network design. In greenfield models, there is no question it is the architecture of choice. It cannot be overstated that any decision to deploy an FTTN overbuild network should include a solution that provides a relatively easy migration to FTTP.
That’s pretty much the long & short of it, be prepared for the future, & the only way to do so is to look at historical data use & predict based on this. ABS, CISCO, & Ookla all show trends toward exponential data growth, yet the assumption is that in the next 6 years data needs will only double (or as Mal says 5x the current average, not the current need). This is clearly incorrect as ABS shows data use is doubling every 18 months or less.
The idea that the average is what people need now is ridiculous, having moved from a 13Mbps DSL service (almost 3x the average) to a 5Mbps DSL service (average in Australia), I can assure you that average is far from what I need in a household of 2. In fact, 13Mbps was pushing it.
So where does Turnbull get his business case from. Well, I think I found where Turnbull lifted 90% of his policy from, but doesn’t reference it once. It’s where all the pretty graphs that show FTTN is cheaper than FTTP come from. I wonder when it’s from. Last year? Nope. Two years ago? Nope. Five years ago, has to be? Nope, Six years ago!
A whitepaper from 2007 by Alcatel-Lucent titled: Deploying Fiber-to-the-Most-Economic Point.
Now, I’m no nuclear physicist, but using six year old data seems like a stretch when arguing economics. In that time we’ve had 8 generations of Intel CPUs, we’ve seen technology drop in price so rapidly that it’s hard to keep track of, & somehow fibre deployments have either stayed the same or gone up in price.
Not buying it.
The document clearly states:
But the economics of FTTN are hard to resist, given cost points that can be 50 percent or less than those of PON. Fiber-to-the-Building stands out as a highly cost efficient solution for multi-dwelling/multi-tenant buildings, with cost points that are up to 30 percent less than FTTN.
If this statement is to be believed, with the knowledge that costs associated with FTTP deployments have dropped by as much as 90%, then the cost of FTTP could be as low as 20% of the cost of FTTN. Sobering news for a Communications Minister hell bent on deploying a copper based network.
The amusing part of the whitepaper is that it assumes 25Mbps services. Could this be where Turnbull got his arbitrary minimum from, a 2007 whitepaper by Alcatel-Lucent arguing a case for their newly finalised ITU-T G.993.2 standard? It’s looking that way.
One thing to note, even as early as 2007 it was clear that pushing VDSL2 into higher speeds becomes more pricey, with the whitepaper adding this caveat to cost graph of 25Mbps services:
It should be noted that the costs increase approximately 25 percent from that above if each subscriber is guaranteed 100 Mb/s.
What’s that? Shortening the loop drives the costs up? Don’t forget, the assumptions in the whitepaper on costs are a maximum loop length of 1.1km, NOT the maximum 400m loop lengths that Turnbull keeps quoting his “100Mbps” services from (NB: Turnbull has never given a clear maximum loop length, just averages). If the assumption is to swap out pillars for nodes, then the loop lengths will be maximum 1.6km, not 1.1, further increasing the cost of even delivering 25Mbps services.
Both documents do state that in more complex scenarios (having to redesign your distribution network as will be required here to get loops under 500m comes to mind) FTTN is NOT the best solution.
In many ways this find is damning evidence of a lazy politician not even bothering to vet his own policy, presenting stale data as fact. Imagine if someone demanded 2007 house prices today, they’d be laughed out of an auction or real estate. Why is it somehow ok to take such stale information on technology that’s constantly reducing in cost has fibre production becomes more efficient, & rollout processes become more efficient.
An interesting thing to note is that the Alcatel-Lucent document is also including the costs of pit & pipe, yet doesn’t include the cost of acquiring copper with pit & pipe for FTTN. This instantly makes this comparison useless in the Australian example, as it’s aimed at service providers who already own copper & want to overbuild fibre (sink new pit & pipe).
This throws into doubt any of the claims in the Liberal party’s broadband policy that they can deliver a cheaper solution on aging copper. At most we’re looking at a band-aid that won’t even outlast the project itself. We’ll start seeing nodes upgraded long before 2019 to FTTP, but by this stage it will be too late, the damage to our digital economy will be irreversible.